Description
System Info
Docker image: tensorrt_llm-release:v0.20.0rc0
GPU: B200
Who can help?
No response
Information
- The official example scripts
- My own modified scripts
Tasks
- An officially supported task in the
examples
folder (such as GLUE/SQuAD, ...) - My own task or dataset (give details below)
Reproduction
YML:
cat dsr1.yml
pytorch_backend_config:
enable_overlap_scheduler: true
use_cuda_graph: true
moe_backend: TRTLLM
Serve:
trtllm-serve serve --tp_size 8 --ep_size 8 --backend pytorch --extra_llm_api_options dsr1.yml /root/.cache/huggingf
ace/hub/models--nvidia--DeepSeek-R1-FP4/snapshots/574fdb8a5347fdbc06b2c18488699c0c17d71e05
Eval
lm_eval --model local-chat-completions --model_args model=/root/.cache/huggingface/hub/models--nvidia--DeepSeek-R1-FP4/snapshots/574fdb8a5347fdbc06b2c18488699c0c17d71e05/,base_url=http://127.0.0.1:8000/v1/chat/completions,num_concurrent=64,timeout=999999,max_gen_toks=8192 --tasks gsm8k --batch_size 64 --num_fewshot 8
Expected behavior
Correctly response the request
actual behavior
API request failed with error message: {"object":"error","message":"[{'type': 'literal_error', 'loc': ('body', 'messages', 0, 'typed-dict', 'role'), 'msg': "Input should be 'developer'", 'input': 'user', 'ctx': {'expected': "'developer'"}}, {'type': 'extra_forbidden', 'loc': ('body', 'messages', 0, 'typed-dict', 'type'), 'msg': 'Extra inputs are not permitted', 'input': 'text'}, {'type': 'literal_error', 'loc': ('body', 'messages', 0, 'typed-dict', 'role'), 'msg': "Input should be 'system'", 'input': 'user', 'ctx': {'expected': "'system'"}}, {'type': 'extra_forbidden', 'loc': ('body', 'messages', 0, 'typed-dict', 'type'), 'msg': 'Extra inputs are not permitted', 'input': 'text'}, {'type': 'extra_forbidden', 'loc': ('body', 'messages', 0, 'typed-dict', 'type'), 'msg': 'Extra inputs are not permitted', 'input': 'text'}, {'type': 'literal_error', 'loc': ('body', 'messages', 0, 'typed-dict', 'role'), 'msg': "Input should be 'assistant'", 'input': 'user', 'ctx': {'expected': "'assistant'"}}, {'type': 'extra_forbidden', 'loc': ('body', 'messages', 0, 'typed-dict', 'type'), 'msg': 'Extra inputs are not permitted', 'input': 'text'}, {'type': 'literal_error', 'loc': ('body', 'messages', 0, 'typed-dict', 'role'), 'msg': "Input should be 'tool'", 'input': 'user', 'ctx': {'expected': "'tool'"}}, {'type': 'missing', 'loc': ('body', 'messages', 0, 'typed-dict', 'tool_call_id'), 'msg': 'Field required', 'input': {'role': 'user', 'content': "Question: Farrah ordered 4 boxes from Amazon containing 20 matchboxes each. If each matchbox has 300 sticks, calculate the total number of match sticks that Farah ordered?\nAnswer: Farrah ordered 4 boxes from Amazon containing 20 matchboxes each, a total of 420=<<420=80>>80 matchboxes.\nIf each matchbox has 300 sticks, the total number of match sticks that Farah bought is 30080 = <<30080=24000>>24000\n#### 24000\n\nQuestion: Maya's organization hosted a weekly farmers' market to raise money for the church choir. They sold broccolis, carrots, spinach, and cauliflowers. After adding together all of their earnings, Maya found out that they had made $380. The organization made $57 from broccoli and the sales of the carrots are twice as much as the sales of broccoli. Then, their sales for the spinach is
additional notes
Ref to the openai api