Skip to content
This repository has been archived by the owner on Oct 11, 2024. It is now read-only.

Commit

Permalink
here we go ...
Browse files Browse the repository at this point in the history
  • Loading branch information
andy-neuma committed Jun 6, 2024
1 parent cbbb67f commit 7723eb7
Show file tree
Hide file tree
Showing 2 changed files with 83 additions and 83 deletions.
106 changes: 53 additions & 53 deletions .github/workflows/build-test.yml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -81,18 +81,18 @@ jobs:
python: ${{ inputs.python }}
secrets: inherit

# TEST-SOLO:
# needs: [BUILD]
# if: success()
# uses: ./.github/workflows/test.yml
# with:
# test_label: ${{ inputs.test_label_solo }}
# timeout: ${{ inputs.test_timeout }}
# gitref: ${{ github.ref }}
# python: ${{ inputs.python }}
# whl: ${{ needs.BUILD.outputs.whl }}
# test_skip_list: ${{ inputs.test_skip_list }}
# secrets: inherit
TEST-SOLO:
needs: [BUILD]
if: success()
uses: ./.github/workflows/test.yml
with:
test_label: ${{ inputs.test_label_solo }}
timeout: ${{ inputs.test_timeout }}
gitref: ${{ github.ref }}
python: ${{ inputs.python }}
whl: ${{ needs.BUILD.outputs.whl }}
test_skip_list: ${{ inputs.test_skip_list }}
secrets: inherit

# TEST-MULTI:
# needs: [BUILD]
Expand All @@ -108,7 +108,7 @@ jobs:
# secrets: inherit

UPLOAD:
needs: [BUILD]
needs: [TEST-SOLO]
if: contains(fromJSON('["NIGHTLY", "WEEKLY", "RELEASE"]'), inputs.wf_category)
uses: ./.github/workflows/nm-upload-assets-to-gcp.yml
with:
Expand All @@ -118,45 +118,45 @@ jobs:
python: ${{ inputs.python }}
secrets: inherit

# BENCHMARK:
# needs: [BUILD]
# if: success()
# uses: ./.github/workflows/nm-benchmark.yml
# with:
# label: ${{ inputs.test_label_solo }}
# benchmark_config_list_file: ${{ inputs.benchmark_config_list_file }}
# timeout: ${{ inputs.benchmark_timeout }}
# gitref: ${{ github.ref }}
# python: ${{ inputs.python }}
# whl: ${{ needs.BUILD.outputs.whl }}
# # Always push if it is a scheduled job
# push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
# secrets: inherit
BENCHMARK:
needs: [BUILD]
if: success()
uses: ./.github/workflows/nm-benchmark.yml
with:
label: ${{ inputs.test_label_solo }}
benchmark_config_list_file: ${{ inputs.benchmark_config_list_file }}
timeout: ${{ inputs.benchmark_timeout }}
gitref: ${{ github.ref }}
python: ${{ inputs.python }}
whl: ${{ needs.BUILD.outputs.whl }}
# Always push if it is a scheduled job
push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
secrets: inherit

# TEST-ACCURACY-SMOKE:
# needs: [BUILD]
# if: inputs.wf_category == 'NIGHTLY'
# uses: ./.github/workflows/nm-test-accuracy-smoke.yml
# with:
# label: ${{ inputs.test_label_solo }}
# timeout: ${{ inputs.benchmark_timeout }}
# gitref: ${{ inputs.gitref }}
# Gi_per_thread: ${{ inputs.Gi_per_thread }}
# nvcc_threads: ${{ inputs.nvcc_threads }}
# python: ${{ inputs.python }}
# whl: ${{ needs.BUILD.outputs.whl }}
# secrets: inherit
TEST-ACCURACY-SMOKE:
needs: [BUILD]
if: inputs.wf_category == 'NIGHTLY'
uses: ./.github/workflows/nm-test-accuracy-smoke.yml
with:
label: ${{ inputs.test_label_solo }}
timeout: ${{ inputs.benchmark_timeout }}
gitref: ${{ inputs.gitref }}
Gi_per_thread: ${{ inputs.Gi_per_thread }}
nvcc_threads: ${{ inputs.nvcc_threads }}
python: ${{ inputs.python }}
whl: ${{ needs.BUILD.outputs.whl }}
secrets: inherit

# TEST-ACCURACY-FULL:
# needs: [BUILD]
# if: ${{ inputs.wf_category == 'WEEKLY' || inputs.wf_category == 'RELEASE' }}
# uses: ./.github/workflows/nm-test-accuracy-full.yml
# with:
# label: ${{ inputs.test_label_multi }}
# timeout: ${{ inputs.benchmark_timeout }}
# gitref: ${{ inputs.gitref }}
# Gi_per_thread: ${{ inputs.Gi_per_thread }}
# nvcc_threads: ${{ inputs.nvcc_threads }}
# python: ${{ inputs.python }}
# whl: ${{ needs.BUILD.outputs.whl }}
# secrets: inherit
TEST-ACCURACY-FULL:
needs: [BUILD]
if: ${{ inputs.wf_category == 'WEEKLY' || inputs.wf_category == 'RELEASE' }}
uses: ./.github/workflows/nm-test-accuracy-full.yml
with:
label: ${{ inputs.test_label_multi }}
timeout: ${{ inputs.benchmark_timeout }}
gitref: ${{ inputs.gitref }}
Gi_per_thread: ${{ inputs.Gi_per_thread }}
nvcc_threads: ${{ inputs.nvcc_threads }}
python: ${{ inputs.python }}
whl: ${{ needs.BUILD.outputs.whl }}
secrets: inherit
60 changes: 30 additions & 30 deletions .github/workflows/nightly.yml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -17,41 +17,41 @@ on:

jobs:

# PYTHON-3-8:
# uses: ./.github/workflows/build-test.yml
# with:
# wf_category: NIGHTLY
# python: 3.8.17
# gitref: ${{ github.ref }}
PYTHON-3-8:
uses: ./.github/workflows/build-test.yml
with:
wf_category: NIGHTLY
python: 3.8.17
gitref: ${{ github.ref }}

# test_label_solo: aws-test-a10g-24G
# test_label_multi: ignore
# test_timeout: 480
# test_skip_list: neuralmagic/tests/skip-for-remote-push-tmp.txt
test_label_solo: aws-test-a10g-24G
test_label_multi: ignore
test_timeout: 480
test_skip_list: neuralmagic/tests/skip-for-remote-push-tmp.txt

# benchmark_label: aws-test-a10g-24G
# benchmark_config_list_file: ./.github/data/nm_benchmark_nightly_configs_list.txt
# benchmark_timeout: 720
# push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
# secrets: inherit
benchmark_label: aws-test-a10g-24G
benchmark_config_list_file: ./.github/data/nm_benchmark_nightly_configs_list.txt
benchmark_timeout: 720
push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
secrets: inherit

# PYTHON-3-9:
# uses: ./.github/workflows/build-test.yml
# with:
# wf_category: NIGHTLY
# python: 3.9.17
# gitref: ${{ github.ref }}
PYTHON-3-9:
uses: ./.github/workflows/build-test.yml
with:
wf_category: NIGHTLY
python: 3.9.17
gitref: ${{ github.ref }}

# test_label_solo: aws-test-a10g-24G
# test_label_multi: ignore
# test_timeout: 480
# test_skip_list: neuralmagic/tests/skip-for-remote-push-tmp.txt
test_label_solo: aws-test-a10g-24G
test_label_multi: ignore
test_timeout: 480
test_skip_list: neuralmagic/tests/skip-for-remote-push-tmp.txt

# benchmark_label: aws-test-a10g-24G
# benchmark_config_list_file: ./.github/data/nm_benchmark_nightly_configs_list.txt
# benchmark_timeout: 720
# push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
# secrets: inherit
benchmark_label: aws-test-a10g-24G
benchmark_config_list_file: ./.github/data/nm_benchmark_nightly_configs_list.txt
benchmark_timeout: 720
push_benchmark_results_to_gh_pages: "${{ github.event_name == 'schedule' || inputs.push_benchmark_results_to_gh_pages }}"
secrets: inherit

PYTHON-3-10:
uses: ./.github/workflows/build-test.yml
Expand Down

8 comments on commit 7723eb7

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

bigger_is_better

Benchmark suite Current: 7723eb7 Previous: 367c5ee Ratio
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.026606741512829 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1303.4588763966678 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.482678296521204 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1420.5159253846043 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1024.5549745484911 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.872109330937107 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1543.3742130218238 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 47.920802638971 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3114.852171533115 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4637383016802372 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.5845826661539 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 115.0194651714103 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.83610920691197 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2188.694196898556 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.884168321944726 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4044.578349157644 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23919626631137011 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.095514620478113 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.23652524871726 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2110.748282333244 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.2610661522272 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3129.677533637309 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.069166571890628 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4139.482451379893 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.88712061034901 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4082.9899968596956 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9098814954031887 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.28459440241454 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 29.09078638067215 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3752.711443106707 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.946421558069044 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3584.230340423744 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9838873617462706 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.102468235427 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.37536665747166 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 14.240167111249825 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3659.722947591205 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.493187025772991 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3844.004944221544 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4919059351219605 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.03041489013904 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.08449435146898 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9088195576717699 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.14654249733009 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.544456335537843 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1123.2027681685872 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 810.2130959152535 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7938734242763164 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 233.20354515592115 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.040473706004068 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4180.930623602335 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.759222279174765 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3853.202836154134 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.3560510098083487 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 727.6113798557463 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 540.7105653496697 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7165022477670971 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 93.14529220972264 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 20.197286201674856 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2625.6472062177313 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7083563284500538 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 557.1587773438247 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 388.28889318075585 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0080573582635406 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4114.509527081995 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.25506998073209264 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 33.15909749517204 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.9797691812162725 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 257.3699935581154 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.727914872937601 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 484.6289334818881 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.873080514873517 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3969.907527745355 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.978113908438463 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3592.375274468685 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.849254461468971 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 587.5648528788438 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 414.07519632167083 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23969679494475696 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.160583342818402 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.6297369641271695 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 861.865805336532 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.01195139593811 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 131.55368147195426 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839418183052333 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.1185249333591 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.2100117599507 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.761501246101922 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3855.53877725447 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.767754616189802 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2696.7713483457037 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.49342128190783 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3844.125117618717 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7951881708947932 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 233.37446221632314 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.536790358201852 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3992.955122057876 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.267802346777064 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1707.407152540509 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.540824514461046 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1438.9418803875608 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1037.938640299163 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.2302204551531797 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 688.7515490967727 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 511.9099351394932 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.134877860967096 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3335.9834926737753 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7465633135567489 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 97.05323076237735 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.4919665945507825 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 843.9556572916017 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.893944960902726 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 376.2128449173544 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.653186798790575 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1384.9142838427747 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.909106054797789 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.18378712371258 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.491894646417626 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.02743083403524 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.0193859903563 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.003579605775197 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3096.4617691450007 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.072994837342242 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4247.566421714255 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.451796696674203 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 757.1802011782385 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 563.5634505730078 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.474786623320928 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.50509600865412 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 99.38866648184761 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 49.98451479632892 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3248.99346176138 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.09826578760934 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3108.6762866016047 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.300257172208379 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1362.7084953011133 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 859.6472102674883 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.041453499878124 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4125.265645437477 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3966631518744186 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 441.5662097436744 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 22.8299050842683 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2945.057755870611 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.854466553962128 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1480.1151567214572 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3943825300640524 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 441.26972890832684 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.063604189599012 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4228.324984488376 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.86773102484081 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1616.4025166146528 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4657709206070573 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.95066721108974 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.02324614235762 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.304223526929059 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 711.6056710764115 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 461.8278407573015 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.76392857622005 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 879.3107149086064 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9474168803779209 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 279.3490252356979 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.51572391091386 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.981543412096745 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3593.256656908863 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9835699336023473 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.0088734890975 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.29373446873066 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 28.585200920920897 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3687.490918798796 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7955130613838493 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 233.4166979799004 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.426037230532863 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1355.3848399692722 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.596161811096142 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1139.5877163182565 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 822.3630906378436 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.238749945359046 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1526.6366067441122 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.772303824955395 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2698.8983763961446 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 20.68373391196801 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2688.8854085558414 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.37483535745853 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 698.728596469609 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.024639775037262 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1563.203170754844 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.662202222227986 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1668.043144444819 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.401513203030136 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 442.19671639391765 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.293801414746696 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1644.0970919585352 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.967980909277437 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 288.6293209253152 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 210.02604448895642 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.577202339374429 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 725.0363041186757 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.16617089236725 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4270.325164676431 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.030138857251492 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1563.918051442694 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0413943119050537 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4182.816945093455 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7581832184748083 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3852.1377989366783 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.76618987341536 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2696.0397372140856 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.341429735401176 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3432.558887084174 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23933032938413334 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.112942819937334 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.781169105857732 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 881.5519837615051 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.028251643405274 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4128.957934490406 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.781034113796474 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 361.5344347935416 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.497574965511481 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3846.2559573073895 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.448019642320522 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 756.0137460723723 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 560.9326927889449 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0414905058447474 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4183.014046475888 tokens/s

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

smaller_is_better

Benchmark suite Current: 7723eb7 Previous: 367c5ee Ratio
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12783.083155500208 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 202.1380538419926 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 157.43865349941188 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 99.7690127711828 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 102.39652851451683 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6391.745147500046 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 110.62699600666822 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 67.78080100002626 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.31836577942342 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 37.02515690245707 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1846.3100610001675 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 90.77630478669259 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 57.26714750016981 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.722670029005814 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.96426746997867 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2059.9623005000467 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 80.68507503328874 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.50467100003152 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.672313126473055 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.80935120898405 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 56187.668163999886 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 34441.326834248 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 34019.45027299985 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 95.89403091889278 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 97.74157201620737 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5293.037485500463 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 182.19029232265407 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 172.6164105002681 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.18959639492836 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.51012748888596 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 74650.93854849988 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 55286.70469496934 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 67708.9426524999 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 66.88689525821377 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 64.41263926529051 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 249498.71426500022 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 235540.36736704665 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 236388.9392779997 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.46762100460508 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 65.61479491696096 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1932.407495499774 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 93.99437012666264 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 59.837363000497135 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.304935880312172 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.585886620835169 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11546.32585950003 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 190.77841928133785 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 148.1468915003461 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 86.58260934045113 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 88.62240363674451 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7613.761868000097 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 154.90966372533998 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.1430349999464 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 56.245646594264336 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 55.84110003784429 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1975.6439519997002 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 78.59668176000923 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.67811850038561 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.211502884317365 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.31649217041812 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2413.6048085001676 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 113.7602833253186 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79.12334050024583 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17.877523090845145 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.1038405826949 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5109.771458000068 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 89.09158235995469 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 61.48387949997414 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.41170657806391 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 30.95482385573757 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12615.810260000217 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 163.39570620000936 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.64198949905403 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 112.02662987045665 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 100.8163297794006 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6111.3240004999625 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 105.71055407334825 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 67.49391649987047 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 34.71013146310299 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.21578957481347 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5747.25785550072 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 115.96191738270136 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 87.37648299938883 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 52.195625971754474 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 45.54364807823351 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6050.696359999961 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.41031202000113 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79.51928649993079 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.495861188214185 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.3000549665603 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3660.5600934990434 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 142.1134080166909 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 92.31942000042181 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.423501766435226 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.73889322522927 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 56700.81794199996 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24098.30956777735 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 22633.730614500564 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 180.90249571215125 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 192.58353225982006 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6060.5851484999675 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 120.68791064668099 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 80.34676250008488 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.58181331325141 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.309595013037224 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2565.8904689998963 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.44517718798306 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 83.49692499996308 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.94325561930678 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.945171436311274 ms

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

bigger_is_better

Benchmark suite Current: 7723eb7 Previous: 367c5ee Ratio
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.474750315128345 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 841.7175409666849 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.0072651534987156 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.94446995483304 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.438562710983845 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3967.710616722848 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.980917898982394 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2467.5193268677112 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0410617371260784 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4182.135499371335 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9042499175540353 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.55248928202458 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.071047903075592 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4243.577153401888 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.329713931103837 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 822.8628110434988 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.632610782880147 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2633.5834976434417 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7600894197274033 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3854.091655220588 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23901941799987345 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.07252433998355 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.01647767997078 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4116.88962197005 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.790986210853696 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4058.2834561893997 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839362988549972 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.1168975050588 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.46792458708526 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 49.04703812588824 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3188.0574781827354 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9059707078811505 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.77619202454957 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.405352561102333 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1651.3479164716518 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0407535004883313 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4181.503922500591 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.556866877851178 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1661.1963470603264 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.135130575745881 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4238.508840139529 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.352195212544904 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.7853776308375 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.53969344020891 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 330.1601472271583 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.6802270877799854 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 547.9847818029971 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 381.90441553847216 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.4142967601088485 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1081.956500310893 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 780.4627154040813 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7927304905572736 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1456.408508373993 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4632634116178392 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.4590502270596 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 114.9016797722006 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.679987595165648 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 478.3983873715342 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.334277088560422 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1373.4890665939122 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 990.9833344923368 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.1893087264688305 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1503.3316971607067 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 14.15176439900463 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3637.00345054419 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.442395521941135 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1487.5114178523477 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7807323118461762 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.49520054000288 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7816333039090884 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.6123295081815 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.3459090640489584 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 724.4792765533597 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 538.3892580779881 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.4458092217956615 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 707.9551988334359 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 9.99288505334515 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1299.0750569348695 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.953245715404364 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3585.984148858922 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 28.866436958348206 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3723.7703676269184 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.479470387108486 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3836.968308586653 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7596165398985963 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3853.6069533960613 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.8081231760488699 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 574.4961821414581 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 404.85083327184896 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.078891567368792 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1695.1279518789713 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.487816629681088 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3841.2499310263984 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.040753479648133 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4181.503879799024 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.121722516323557 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1306.1326481977721 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 822.4430152771808 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7809254455656294 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.52030792353185 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9826238818768152 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 289.7299269843185 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.08953866429178 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.02742128579661 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4118.067119613661 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.481315337951703 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 842.5709939337213 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.320928088051042 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1471.7206514466357 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9059711387923123 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.7762480430006 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.4195314383637916 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1083.615317503102 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 781.7094461852139 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.87881537517926 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3566.85555142107 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.384768036422322 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3145.635076698479 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.2205907316570515 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 685.7776336885416 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 509.6877494557797 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.9634133670531637 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 255.24373771691128 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.017520073752533 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3281.1116856837343 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.35525638000082 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 436.1833294001066 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.056911053962844 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1307.3984370151697 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.588014837631489 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 856.4419288920936 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.998950957774394 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3095.8646735528964 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.49190212559007307 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.02940787847993 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.01468257886566 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.004161922309827 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2080.5410499002774 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.46540607436049447 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.8520079214025 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.9307965357869 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.4432301073410847 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 754.5346099497893 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 559.0273367603558 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.863805268991468 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3960.400400716255 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.229216059850909 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1340.1962772061545 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 966.6719148001223 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.044377561310099 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4188.929623124393 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.888424431600278 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3081.606751676436 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.2388947407536361 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.056316297972693 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.202414442462697 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2496.3138775201505 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.844564218549479 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 369.7933484114323 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.362863608824004 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3142.8094055382962 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.57705462335847 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2155.017101036601 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.967628188814441 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 288.52414791339396 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 209.95273898166135 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7122551146364847 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 92.59316490274301 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7426137766104906 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 96.53979095936377 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 27.259719679113136 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3516.5038386055944 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.4477194578405777 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 755.9210411000461 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 561.8723697343327 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7641274856197713 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3858.2306727602654 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.2546443374974644 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 33.10376387467037 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.286021113934051 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1467.1827448114268 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.623015710868888 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2629.097225773857 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0053800375865025 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4109.023697014744 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 47.51756142014626 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3088.6414923095067 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.87781381495802 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3566.598150444211 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.974463830702835 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1688.3401489956843 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23900793119478395 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.071031055321914 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4745374315856206 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.43922466534295 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 99.33650234525658 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.248380385728818 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 682.2894501447463 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3522140806324243 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.78783048221516 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4919130854983481 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.03230502063334 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.72553271685123 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.628080326621224 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2631.46523751502 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.292221234114496 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 707.8990429941327 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 458.853790350061 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 9.634811052148212 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1252.5254367792675 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9466216325351265 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 279.11454375842385 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.344056409916 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839263455852075 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.11396275028375 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.38706120353763 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.479001562541648 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3836.7278015838656 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.8279845483389225 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4015.7560732978673 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.838507449726522 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3197.4125431941325 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.0380218367886 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4123.505202272551 tokens/s

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

smaller_is_better

Benchmark suite Current: 7723eb7 Previous: 367c5ee Ratio
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1879.7467824997511 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 91.39874358668749 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.1986459997097 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.988480969677605 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.187323198428148 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 78176.76265650016 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58714.22578044001 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 71458.93627100009 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.70006333325242 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 66.01555423348927 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 65131.474450500034 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 42288.03893590467 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 42407.38135850006 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 100.31667559752744 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 102.74020092448394 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6429.881593499999 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 109.54886699999938 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 70.86481550004464 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.68720643219057 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 37.37856545948284 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15799.322397000196 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 211.9199030946893 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 169.4760860000315 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 129.27597515334028 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 134.67790265669265 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5465.411713499634 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 183.27851867063387 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 173.93126100068912 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.25995555596707 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 38.509064551247256 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 258815.88115349997 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 244499.44135369203 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 245863.29692499977 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 70.30434243411553 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 67.27276355737541 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18799.036059999708 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1491.5206421146625 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 270.1706700008799 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 157.39919167075823 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 146.01257559617974 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3718.5346094993292 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 144.20874462672145 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 97.92165949966147 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.68959320588485 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.129002200571332 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 70694.81159749921 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31557.407188473342 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 30425.266239999473 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 193.56484348518538 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.4877101088725 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7957.695361499873 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 159.35679767599777 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 132.67411349988834 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.90024433355537 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.53508037248212 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1995.1480949998768 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 78.27047056666439 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.84506050003256 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.44132130489565 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.510436644360468 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6226.167451499805 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 107.4018972066339 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.34590850007771 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.0631127635019 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.60291497675336 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2644.836326999666 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 120.56133040932764 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 85.77099650028686 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.944027429523835 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17.830635445205655 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18172.08555649995 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 227.4976878459996 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 179.31617149997692 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 150.08463046700945 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 154.8834398367518 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6116.362859499986 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.79129089333553 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79.58853550007916 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.2272690743536 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.82451299031622 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2515.1216479998766 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 115.89628272133632 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 80.1281165004184 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.64714923920508 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.942370406302935 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5222.246493500279 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 90.68277033334502 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 62.94527600039146 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.79760600398387 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.32337244028812 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2072.7505139998357 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 80.23777848665607 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.41608849977274 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.936985607353826 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.968076143964355 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6086.7290565001895 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 119.0172619653531 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 88.82956800061947 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 55.029380314578795 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 48.10261243347492 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6130.04963100002 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 127.78770609333681 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 84.37591050005722 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.08508923996274 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.865117427732756 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1965.8358245001182 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 94.72450298669477 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 59.38587799982997 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.52438462878134 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.829621100778368 ms

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'bigger_is_better'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.10.

Benchmark suite Current: 7723eb7 Previous: 367c5ee Ratio
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 21.433299156481482 prompts/s 24.347844802232547 prompts/s 1.14
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2764.895591186111 tokens/s 3140.8719794879985 tokens/s 1.14

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

bigger_is_better

Benchmark suite Current: 7723eb7 Previous: 367c5ee Ratio
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839626513087448 prompts/s 0.9839101054644096 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.12466761388777 tokens/s 290.10917429653273 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.38177853398386 tokens/s 212.59673618804652 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.872758338199382 prompts/s 7.869930572422599 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4038.7250274962835 tokens/s 4037.2743836527934 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.091151312255519 prompts/s 7.108328445960933 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3315.5387075581907 tokens/s 3323.570048193494 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0598663106737307 prompts/s 2.0613536433530886 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4220.666070570474 tokens/s 4223.713615230478 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9838919296440267 prompts/s 0.9839597742950889 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.10381509530674 tokens/s 290.12381931682125 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.5665717598265 tokens/s 212.4008367452522 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7583392238551685 prompts/s 3.754339255944596 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3852.297704451548 tokens/s 3848.197737343211 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7792096342925554 prompts/s 1.779135225535058 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.2972524580322 tokens/s 231.28757931955755 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.476031087881767 prompts/s 6.456841927191974 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 841.8840414246296 tokens/s 839.3894505349566 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.439678298872166 prompts/s 3.4194354621572383 prompts/s 0.99
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1089.9996561296007 tokens/s 1083.5849036030072 tokens/s 0.99
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 786.3425627862999 tokens/s 781.6669892836545 tokens/s 0.99
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.163666918103562 prompts/s 4.159574591992457 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4267.758591056151 tokens/s 4263.563956792269 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.929512959310234 prompts/s 13.921121699644356 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3579.88483054273 tokens/s 3577.7282768085993 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4919176893337273 prompts/s 0.49191243467387813 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.03352199847748 tokens/s 130.03213298169294 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.07757323632 tokens/s 121.9844455504283 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0053824062804417 prompts/s 2.004013160594993 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4109.028550468625 tokens/s 4106.22296605914 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.864489331975435 prompts/s 3.8639718090630004 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3961.101565274821 tokens/s 3960.5711042895755 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.7335374465296836 prompts/s 2.720702713763779 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 355.35986804885886 tokens/s 353.6913527892913 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.262974076777441 prompts/s 11.305613118664521 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1464.1866299810672 tokens/s 1469.7297054263877 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.8143646270494473 prompts/s 1.812925996749661 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 576.479282528839 tokens/s 576.0221855646009 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 406.24349745487945 tokens/s 405.9358857842101 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.758219192444273 prompts/s 3.754728348201599 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3852.17467225538 tokens/s 3848.5965569066393 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.096813665884932 prompts/s 8.091113380492665 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4153.66541059897 tokens/s 4150.741164192737 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.120385433570401 prompts/s 4.1249435234677385 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1305.7089400441243 tokens/s 1307.1533531516916 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 822.1405055603021 tokens/s 822.5962374499364 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9464081578987101 prompts/s 0.9467960871672213 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 279.0516000502943 tokens/s 279.16598228821243 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.2979743517349 tokens/s 204.3785593628736 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.8480386195633858 prompts/s 2.8475734056852877 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 370.2450205432401 tokens/s 370.1845427390874 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 9.60887525492067 prompts/s 9.61022686900094 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1249.1537831396872 tokens/s 1249.3294929701221 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.477209795158185 prompts/s 25.301403603475283 prompts/s 0.99
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1656.018636685282 tokens/s 1644.5912342258935 tokens/s 0.99
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.4475132031296276 prompts/s 2.4484855408375745 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 755.8573441451791 tokens/s 756.1576279583986 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 562.2623131285632 tokens/s 562.5836257471678 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.905243059897921 prompts/s 0.9059123416639003 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.68159778672972 tokens/s 117.76860441630704 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.781475549758745 prompts/s 1.7818510691275493 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.59182146863685 tokens/s 231.64063898658142 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.347877306076514 prompts/s 3.3493245276862114 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.22404978994683 tokens/s 435.41218859920747 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.2401727903711945 prompts/s 5.2551436071234 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 681.2224627482552 tokens/s 683.168668926042 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.015409528705026 prompts/s 8.016524365387166 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4111.905088225679 tokens/s 4112.476999443616 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0387548208605186 prompts/s 2.0365089770478697 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4177.408627943203 tokens/s 4172.806893971085 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.311220227342883 prompts/s 11.299495679250603 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1470.4586295545746 tokens/s 1468.9344383025784 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.211745821253064 prompts/s 19.176074537220124 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2497.5269567628984 tokens/s 2492.889689838616 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.43900903407262 prompts/s 11.479895351113106 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1487.0711744294408 tokens/s 1492.3863956447037 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.630090700018801 prompts/s 5.639548764506602 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2632.405207700791 tokens/s 2636.827420332707 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.461256402659944 prompts/s 6.469168306125793 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 839.9633323457928 tokens/s 840.991879796353 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.46553622936506356 prompts/s 0.4655402239127167 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.88720357003805 tokens/s 125.88828374898411 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.96377694624466 tokens/s 117.96478913798967 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.282534782113451 prompts/s 4.233958128721013 prompts/s 0.99
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1357.0924471039316 tokens/s 1341.6989914104017 tokens/s 0.99
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 978.0224533535654 tokens/s 967.4735456065139 tokens/s 0.99
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23879046539175106 prompts/s 0.23864663250332274 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.04276050092764 tokens/s 31.02406222543196 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.2948283955373228 prompts/s 2.2936273348371725 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 708.7042039658062 tokens/s 708.3332843933144 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 459.4062861594077 tokens/s 460.3065407435822 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4745958256374516 prompts/s 0.47463484376330717 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.45466054900396 tokens/s 125.46497460039262 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 99.3487261667732 tokens/s 99.35689396111897 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.6695527191068735 prompts/s 1.6752285596489436 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 544.5034716689314 tokens/s 546.3545751677474 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 379.47820270153164 tokens/s 380.76604977972767 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.420846796657735 prompts/s 23.94916741240376 prompts/s 0.98
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3150.289236768848 tokens/s 3089.442596200085 tokens/s 0.98
{"name": "request_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.632150745611533 prompts/s 5.632220671950128 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2633.3684026181286 tokens/s 2633.401097377002 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.330254102886536 prompts/s 4.349577679450244 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1372.2142226637143 tokens/s 1378.3376708409876 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 989.4139862964073 tokens/s 994.5309364062982 tokens/s 1.01
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0711891358970025 prompts/s 2.0700109241609885 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4243.866539452958 tokens/s 4241.452383605866 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.990582857921666 prompts/s 16.03671526578408 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2078.7757715298167 tokens/s 2084.77298455193 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4919106297968047 prompts/s 0.4919119148124554 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.03165588048736 tokens/s 130.03199556152447 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.00367440220352 tokens/s 122.57133152020222 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.345805762171402 prompts/s 2.347024053770863 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 724.4473741788535 tokens/s 724.823615112543 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 538.365550159353 tokens/s 538.638890975008 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.909267375947926 prompts/s 6.917690413777854 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3230.4970542982123 tokens/s 3234.4353298659735 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.027993766454937 prompts/s 24.091661255705365 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3099.611195872687 tokens/s 3107.8243019859924 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.469541264902457 prompts/s 7.456689608173404 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3831.8746688949605 tokens/s 3825.2817689929566 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3473851274908135 prompts/s 3.35200325893339 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.16006657380575 tokens/s 435.7604236613407 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.221534231014292 prompts/s 2.222106013781002 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 686.0690114500403 tokens/s 686.2455932159409 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 509.91023306753107 tokens/s 510.04443755518093 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.25475580791678726 prompts/s 0.25458760778218065 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 33.11825502918234 tokens/s 33.096389011683485 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.678235027098695 prompts/s 3.676426423812059 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 478.1705535228304 tokens/s 477.9354350955677 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 28.61815957273523 prompts/s 28.614750487973723 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3691.7425848828443 tokens/s 3691.30281294861 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.791636359789746 prompts/s 3.7921983518136506 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1455.9883621592624 tokens/s 1456.2041670964418 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.440036038935864 prompts/s 5.456431240585442 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 707.2046850616624 tokens/s 709.3360612761073 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.925783705472753 prompts/s 13.866006704075216 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3578.9264123064977 tokens/s 3563.563722947331 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.108152709016164 prompts/s 25.19399502174627 prompts/s 0.96
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1697.0299260860506 tokens/s 1637.6096764135075 tokens/s 0.96
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.962703488973165 prompts/s 1.9629535472290662 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 255.15145356651144 tokens/s 255.18396113977863 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.933747986042652 prompts/s 15.88790182805181 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4094.9732324129614 tokens/s 4083.190769809315 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.0075283114912283 prompts/s 1.0083076041504455 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.97868049385968 tokens/s 131.07998853955792 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.591399915945219 prompts/s 6.592224684944499 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 856.8819890728785 tokens/s 856.989209042785 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4633136761779577 prompts/s 0.46340960307834844 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.47233716088134 tokens/s 122.49769447773062 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 114.91414672349826 tokens/s 114.93793915284583 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.573294472343672 prompts/s 15.544075357682487 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4002.3366793923237 tokens/s 3994.8273669243995 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9677846153067616 prompts/s 0.9677164275152352 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 288.5707906434525 tokens/s 288.5504586350678 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 209.98667988054345 tokens/s 209.96865899940565 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.0779578556045 prompts/s 10.070611890671955 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1310.134521228585 tokens/s 1309.179545787354 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.442486099469433 prompts/s 2.444789686811601 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 754.3048404788133 tokens/s 755.0162496790708 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 560.2639747925638 tokens/s 560.9781818162076 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 21.433299156481482 prompts/s 24.347844802232547 prompts/s 1.14
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2764.895591186111 tokens/s 3140.8719794879985 tokens/s 1.14
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.64671952519404 prompts/s 23.978666468821157 prompts/s 1.01
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3050.426818750031 tokens/s 3093.2479744779293 tokens/s 1.01
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.046637677280396 prompts/s 4.044494519615079 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4147.8036192124055 tokens/s 4145.606882605456 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.039309464119207 prompts/s 2.0351946371431104 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4178.545091980254 tokens/s 4170.113811506233 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0402314421414007 prompts/s 2.0353544942949595 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4180.43422494773 tokens/s 4170.441358810373 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 29.101900509372797 prompts/s 29.103148922260967 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3754.145165709091 tokens/s 3754.3062109716648 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7416501774597816 prompts/s 0.7414713481606056 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 96.41452306977159 tokens/s 96.39127526087873 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7118071448630923 prompts/s 0.712602671481504 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 92.534928832202 tokens/s 92.63834729259551 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9049438582467028 prompts/s 0.9033709577949868 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.64270157207136 tokens/s 117.43822451334827 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.01904792627166 prompts/s 18.925975723929728 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2472.4762304153155 tokens/s 2460.3768441108646 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.644771258978796 prompts/s 16.653812681567196 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2163.8202636672436 tokens/s 2164.9956486037354 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9826518314377993 prompts/s 0.9826575384453134 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 289.73816800553993 tokens/s 289.73985073572885 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.0955712975346 tokens/s 212.09680309803645 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7569804415100667 prompts/s 3.7523887800441478 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3850.9049525478185 tokens/s 3846.1984995452513 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 14.167064513737506 prompts/s 14.156753141096177 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3640.935580030539 tokens/s 3638.2855572617177 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.406749681221683 prompts/s 3.406108885355436 prompts/s 1.00
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1079.5649064823392 tokens/s 1079.361844680284 tokens/s 1.00
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 778.7807059608226 tokens/s 778.5729104930593 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3517859891715394 prompts/s 3.3544893716938997 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.7321785923001 tokens/s 436.08361832020694 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 49.11347718073741 prompts/s 49.4053766498384 prompts/s 1.01
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3192.3760167479318 tokens/s 3211.3494822394964 tokens/s 1.01
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23429117012517067 prompts/s 0.23875898426274353 prompts/s 1.02
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 30.45785211627219 tokens/s 31.03866795415666 tokens/s 1.02
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.479681184313159 prompts/s 7.471103657611703 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3837.0764475526503 tokens/s 3832.676176354804 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9059942943078962 prompts/s 0.9060074242205314 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.77925826002651 tokens/s 117.78096514866908 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23868506161650155 prompts/s 0.2387678542468053 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.0290580101452 tokens/s 31.03982105208469 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.32642134107012 prompts/s 6.33940359735184 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 822.4347743391157 tokens/s 824.1224676557392 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.924548303385965 prompts/s 13.890356057386485 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3578.608913970193 tokens/s 3569.8215067483266 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.191362487813279 prompts/s 3.194667982860906 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1504.299770430594 tokens/s 1505.8578684092138 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.005529141152103 prompts/s 26.0721677091497 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1690.3593941748866 tokens/s 1694.6909010947304 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.992709627006914 prompts/s 25.988072360785914 prompts/s 1.04
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1624.5261257554496 tokens/s 1689.2247034510845 tokens/s 1.04
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.481353325462274 prompts/s 7.477191928896954 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3837.934255962147 tokens/s 3835.7994595241375 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 47.80499082714928 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3107.324403764703 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.6229484768634475 prompts/s 5.629923453346251 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2629.0657898422737 tokens/s 2632.327009846573 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 9.966545030446134 prompts/s 9.97747419947387 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1295.6508539579975 tokens/s 1297.071645931603 tokens/s 1.00
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.780655059939717 prompts/s 1.7806231673027537 prompts/s 1.00
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.48515779216322 tokens/s 231.48101174935798 tokens/s 1.00

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

smaller_is_better

Benchmark suite Current: 7723eb7 Previous: 367c5ee Ratio
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1968.6839945002248 ms 1956.2861680001333 ms 1.01
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 94.97119854666683 ms 95.14779560665677 ms 1.00
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 62.33603250075248 ms 59.077937999518326 ms 1.06
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.651372694325214 ms 13.437431639448432 ms 1.02
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.860478887897477 ms 12.723519856814338 ms 1.01
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1879.242770000019 ms 1889.1761065005994 ms 0.99
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 91.76566043001003 ms 92.37879568333179 ms 0.99
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 59.09067149968905 ms 57.173135000084585 ms 1.03
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.954178463263185 ms 13.024526196932513 ms 0.99
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.196996463373345 ms 12.257026714855618 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 69041.27056349989 ms 70516.79164049984 ms 0.98
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 30786.065317992634 ms 31495.424776235326 ms 0.98
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 30204.45853899946 ms 30031.702047999715 ms 1.01
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 192.07372248916724 ms 193.22762823587541 ms 0.99
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 203.00327425415747 ms 204.57441203650677 ms 0.99
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2073.363577500004 ms 2064.670552500047 ms 1.00
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 80.26193202663003 ms 80.15265723332655 ms 1.00
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.03832299986243 ms 42.677035499764315 ms 0.94
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.851787315132494 ms 11.785276597867119 ms 1.01
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.946649315251008 ms 11.899947824989589 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 258646.48230000012 ms 259011.95400250025 ms 1.00
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 244166.068009794 ms 244055.78354314202 ms 1.00
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 245446.00923049985 ms 245917.08374050018 ms 1.00
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 69.77034353002637 ms 70.01394674477102 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 67.4934526452921 ms 67.20731673968393 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18633.95326800037 ms 18247.279312500723 ms 1.02
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1509.6453839919814 ms 1292.66038330267 ms 1.17
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 271.1712990003434 ms 246.61586699949112 ms 1.10
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 157.6795801950956 ms 156.7642815834863 ms 1.01
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 146.2692175953914 ms 144.93258284940498 ms 1.01
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6181.886618000021 ms 6116.3640780000605 ms 1.01
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 128.63030907000166 ms 129.47920948666857 ms 0.99
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79.32966350006154 ms 92.83823349994691 ms 0.85
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.095042180937796 ms 40.796792863182176 ms 1.01
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.96171586975987 ms 40.748115254378284 ms 1.01
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2516.7066469998645 ms 2491.7438265001692 ms 1.01
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 114.85854687332176 ms 114.97949669731922 ms 1.00
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 80.3256735002833 ms 79.77798499996425 ms 1.01
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.621763581235275 ms 18.55062783805633 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.84339978024299 ms 16.703622409928943 ms 1.01
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6218.226435000133 ms 6185.281503999931 ms 1.01
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 104.84746503999001 ms 106.85994023998623 ms 0.98
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 70.71776900011173 ms 68.0730799997491 ms 1.04
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.0113103722989 ms 34.97793281004091 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.46113156838068 ms 35.46808400963367 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16854.213983499903 ms 18063.48664699999 ms 0.93
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 223.60520914666085 ms 232.45559611600157 ms 0.96
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 179.99765800004752 ms 182.47827300001518 ms 0.99
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 139.72413358499165 ms 149.0605816427107 ms 0.94
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 145.57145586710212 ms 153.05545275452593 ms 0.95
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6041.127196499474 ms 6035.800041500806 ms 1.00
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 119.31926225068796 ms 119.20267250667773 ms 1.00
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 89.68732700031978 ms 87.79566750035883 ms 1.02
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 54.37035866252607 ms 54.44995107384843 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 47.53604664623556 ms 47.61603700214803 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5165.030997500253 ms 5162.35110100024 ms 1.00
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 87.9559120334064 ms 88.73062119988997 ms 0.99
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 60.17887549933221 ms 61.62037499962025 ms 0.98
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.78582089200521 ms 36.810021177324366 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.28434066088498 ms 31.314041988770864 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79797.74271449992 ms 78778.1954510001 ms 1.01
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 59881.56754889332 ms 59162.487471686676 ms 1.01
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 73495.63050949996 ms 72037.40187199991 ms 1.02
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.41308805145677 ms 68.62343158062342 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 65.93914176544594 ms 66.03294767678214 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15897.03651400032 ms 15547.739225999976 ms 1.02
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 211.5834134293394 ms 209.84569314132264 ms 1.01
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 168.54961149965675 ms 169.8408844995356 ms 0.99
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 129.60051243318145 ms 125.59957043409153 ms 1.03
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 135.88059189291783 ms 131.1312142104596 ms 1.04
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1992.1229054998548 ms 1985.349613499693 ms 1.00
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 77.90749754668772 ms 78.35227234002257 ms 0.99
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 38.74097550033184 ms 39.155220500106225 ms 0.99
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.415593659546616 ms 11.33685106604449 ms 1.01
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.485650890122026 ms 11.407172921438526 ms 1.01
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5459.278260500469 ms 5444.995677999941 ms 1.00
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 183.46024677601721 ms 182.43153263602773 ms 1.01
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 177.06370550058637 ms 173.46796700076084 ms 1.02
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.40760773688081 ms 41.280124068436784 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 38.46967377789887 ms 38.37203757283457 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8000.711440000032 ms 7956.53410649993 ms 1.01
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 157.6579761373323 ms 157.04044950666443 ms 1.00
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.48854350006422 ms 131.30081100007374 ms 0.99
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.73946983481814 ms 58.715331306465714 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.40785018415126 ms 58.409235365153606 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6425.155506500005 ms 6432.053577000033 ms 1.00
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 109.76755281335423 ms 111.57128104666981 ms 0.98
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 69.65166550003232 ms 71.33699750005462 ms 0.98
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.688677362807084 ms 36.59753883390603 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 37.366336272395884 ms 37.31719864656108 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6109.539306999977 ms 6106.5911880000385 ms 1.00
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.61916720332616 ms 122.00355583999226 ms 1.01
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 78.28700400000344 ms 83.82650199996533 ms 0.93
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.115151653379435 ms 40.20168109883727 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.78568133483568 ms 39.8089047256667 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2637.2123574997204 ms 2619.209377000061 ms 1.01
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 119.86357280132506 ms 119.57118906267958 ms 1.00
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 84.31642149980689 ms 83.23856800006979 ms 1.01
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.852726088530762 ms 19.589286586639965 ms 1.01
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17.93012695026843 ms 17.653235870041396 ms 1.02
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3702.4795790002827 ms 3704.064616999858 ms 1.00
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 144.78014188669476 ms 143.43571817667907 ms 1.01
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 90.3476210005465 ms 97.69973150014266 ms 0.92
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.671557033121047 ms 24.87243340627162 ms 0.99
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.10726477000276 ms 24.099637962123776 ms 1.00
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 64908.61242949995 ms 65373.15701099999 ms 0.99
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 42041.117765918 ms 42277.50111589267 ms 0.99
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41626.18072700002 ms 41742.67372500003 ms 1.00
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 100.28372351107903 ms 100.17157111138543 ms 1.00
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 102.74819537850958 ms 102.43454124109734 ms 1.00

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'smaller_is_better'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.10.

Benchmark suite Current: 7723eb7 Previous: 367c5ee Ratio
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1509.6453839919814 ms 1292.66038330267 ms 1.17

This comment was automatically generated by workflow using github-action-benchmark.

Please sign in to comment.