Skip to content

GitLab

Commit 44e87dde authored by yangql's avatar yangql
Browse files

修复int4选取config的bug

parent b3312eec
Hide whitespace changes
Inline Side-by-side
Showing with 4 additions and 4 deletions
vllm/model_executor/layers/quantization/utils/fused_moe_cuda.py
View file @ 44e87dde
...@@ -201,11 +201,11 @@ def config_cuda(M): ...@@ -201,11 +201,11 @@ def config_cuda(M):
NearestM = min(reference_points, key=lambda x: abs(x - M)) NearestM = min(reference_points, key=lambda x: abs(x - M))
if device_name == "K100_AI": if device_name == "K100_AI":
mode_1 = k100ai_gemm1_m_to_mode_dict.get(M, k100ai_gemm1_m_to_mode_dict[NearestM]) mode_1 = k100ai_gemm1_m_to_mode_dict.get(NearestM, k100ai_gemm1_m_to_mode_dict[32])
mode_2 = k100ai_gemm2_m_to_mode_dict.get(M, k100ai_gemm2_m_to_mode_dict[NearestM]) mode_2 = k100ai_gemm2_m_to_mode_dict.get(NearestM, k100ai_gemm2_m_to_mode_dict[32])
else: else:
mode_1 = bw_gemm1_m_to_mode_dict.get(M, k100ai_gemm1_m_to_mode_dict[NearestM]) mode_1 = bw_gemm1_m_to_mode_dict.get(NearestM, bw_gemm1_m_to_mode_dict[32])
mode_2 = bw_gemm2_m_to_mode_dict.get(M, k100ai_gemm2_m_to_mode_dict[NearestM]) mode_2 = bw_gemm2_m_to_mode_dict.get(NearestM, bw_gemm2_m_to_mode_dict[32])
return mode_1, mode_2 return mode_1, mode_2
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment