Add directory structure samples

Author: pgmpablo157321

tools/submission/directory_structure_samples/sample_1/closed/AMD/code/llama2-70b-99.9/README.md

@@ -0,0 +1,3 @@
+# Sample README
+
+Use this folder to include your submission code

tools/submission/directory_structure_samples/sample_1/closed/AMD/code/llama2-70b-99/README.md

@@ -0,0 +1,3 @@
+# Sample README
+
+Use this folder to include your submission code

tools/submission/directory_structure_samples/sample_1/closed/AMD/code/mixtral-8x7b/README.md

@@ -0,0 +1,3 @@
+# Sample README
+
+Use this folder to include your submission code

tools/submission/directory_structure_samples/sample_1/closed/AMD/code/stable-diffusion-xl/README.md

@@ -0,0 +1,3 @@
+# Sample README
+
+Use this folder to include your submission code

tools/submission/directory_structure_samples/sample_1/closed/AMD/documentation/README.md

@@ -0,0 +1,3 @@
+# Sample README
+
+Use this folder to include your submission documentation

tools/submission/directory_structure_samples/sample_1/closed/AMD/documentation/calibration.md

@@ -0,0 +1,3 @@
+# Sample README
+
+Use this file to include your submission calibration

tools/submission/directory_structure_samples/sample_1/closed/AMD/results/8xMI300X_2xEPYC_9575F/llama2-70b-99.9/Interactive/README.md

@@ -0,0 +1,94 @@
+# MLPerf Inference 5.1
+
+## Setup
+
+### Model and Dataset
+
+Build the docker image for the benchmark by running the below command
+
+```bash
+bash setup/build_model_and_dataset_env.sh
+```
+
+Start the docker container for the benchmark by running the below command
+
+```bash
+bash setup/start_model_and_dataset_env.sh
+```
+
+Inside the docker, download the model with
+
+```bash
+# Generate an access token on huggingface and set it here
+HUGGINGFACE_ACCESS_TOKEN="<your HF token goes here>" python download_model.py
+```
+
+Inside the docker, download the dataset with
+
+```bash
+bash download_llama2_70b.sh
+```
+
+Inside the docker, quantize the model with
+
+```bash
+bash quantize_llama2_70b.sh
+```
+
+Exit the docker image, because a different image is needed for inference
+
+## Inference
+
+### Runtime tunables
+
+To boost the machine's performance further, execute the following script before any performance test (should be set once after a reboot):
+
+```bash
+bash setup/runtime_tunables.sh
+```
+
+### Docker
+
+```bash
+export MLPERF_IMAGE_NAME=rocm/mlperf-inference:submission_5.1-llama2_70b
+```
+
+Build the docker image for the benchmark by running the below command
+
+```bash
+bash setup/build_submission_llama2_70b.sh $MLPERF_IMAGE_NAME
+```
+
+Start the docker container for the benchmark by running the below command
+
+```bash
+bash setup/start_submission_env.sh $MLPERF_IMAGE_NAME
+```
+
+### Running the benchmark
+
+Run the following commands inside the docker container
+
+``` bash
+## Performance
+python /lab-mlperf-inference/code/llama2-70b-99.9/main.py \
+   --config-path /lab-mlperf-inference/code/llama2-70b-99.9/harness_llm/models/llama2-70b/ \
+   --config-name interactive_mi300x \
+   test_mode=performance \
+   harness_config.device_count=8 \
+   harness_config.user_conf_path=/lab-mlperf-inference/code/llama2-70b-99.9/user_mi300x.conf \
+   harness_config.output_log_dir=/lab-mlperf-inference/results/llama2-70b/Interactive/performance/run_1
+
+## Accuracy
+python /lab-mlperf-inference/code/llama2-70b-99.9/main.py \
+   --config-path /lab-mlperf-inference/code/llama2-70b-99.9/harness_llm/models/llama2-70b/ \
+   --config-name interactive_mi300x \
+   test_mode=accuracy \
+   harness_config.device_count=8 \
+   harness_config.user_conf_path=/lab-mlperf-inference/code/llama2-70b-99.9/user_mi300x.conf \
+   harness_config.output_log_dir=/lab-mlperf-inference/results/llama2-70b/Interactive/accuracy
+
+### Evaluate accuracy
+bash /lab-mlperf-inference/code/llama2-70b-99.9/scripts/check_llama2_accuracy_scores.sh \
+   /lab-mlperf-inference/results/llama2-70b/Interactive/accuracy/mlperf_log_accuracy.json
+```

tools/submission/directory_structure_samples/sample_1/closed/AMD/results/8xMI300X_2xEPYC_9575F/llama2-70b-99.9/Interactive/TEST06/accuracy/mlperf_log_accuracy.json

@@ -0,0 +1,4 @@
+First token check pass: True
+EOS check pass: True
+Sample length check pass: True
+TEST06 verification complete

tools/submission/directory_structure_samples/sample_1/closed/AMD/results/8xMI300X_2xEPYC_9575F/llama2-70b-99.9/Interactive/TEST06/verify_accuracy.txt

@@ -0,0 +1,6 @@
+
+Results
+
+{'rouge1': 44.4542, 'rouge2': 22.0419, 'rougeL': 28.6112, 'rougeLsum': 42.0341, 'gen_len': 29579723, 'gen_num': 24576, 'gen_tok_len': 7399913, 'tokens_per_sample': 301.1}
+
+hash=8f6a1812b3b71c9e53e99a9f7d73df2a2179ca2aac62ef1ddda8400d744cfac6