linzhao-amd commited on
Commit
a94dc97
·
verified ·
1 Parent(s): b2a2844

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -1
README.md CHANGED
@@ -112,7 +112,7 @@ The model was evaluated on AIME24, GPQA Diamond, MATH-500, and GSM8K benchmarks.
112
  ### Reproduction
113
 
114
  The results can be reproduced using vLLM emulation docker: `rocmshared/pytorch:vllm-gfx950-mxfp4-mxfp6-v3`.
115
- The results of AIME24, MATH-500, and GPQA Diamond, were obtained using [vLLM](https://docs.vllm.ai/en/latest/) while GSM8K was obtained using [SGLang](https://docs.sglang.ai/). For AIME24, MATH-500, and GPQA Diamond, we took 10 rounds with different random seeds for reliable performance estimation.
116
 
117
  ```
118
  MODEL_ARGS="model_name=amd/DeepSeek-R1-0528-MXFP4-ASQ,dtype=bfloat16,tensor_parallel_size=8,max_model_length=71536,max_num_batched_tokens=32768,gpu_memory_utilization=0.85,generation_parameters={max_new_tokens:65536,temperature:0.6,top_p:0.95,seed:$SEED}"
 
112
  ### Reproduction
113
 
114
  The results can be reproduced using vLLM emulation docker: `rocmshared/pytorch:vllm-gfx950-mxfp4-mxfp6-v3`.
115
+ The results of AIME24, MATH-500, and GPQA Diamond, were obtained using [vLLM](https://docs.vllm.ai/en/latest/) while GSM8K was obtained using [SGLang](https://docs.sglang.ai/). For AIME24, MATH-500, and GPQA Diamond, we took 10 rounds with different random seeds for reliable performance.
116
 
117
  ```
118
  MODEL_ARGS="model_name=amd/DeepSeek-R1-0528-MXFP4-ASQ,dtype=bfloat16,tensor_parallel_size=8,max_model_length=71536,max_num_batched_tokens=32768,gpu_memory_utilization=0.85,generation_parameters={max_new_tokens:65536,temperature:0.6,top_p:0.95,seed:$SEED}"