[Bugfix] DeepSeek V4: support transformers >= 4.57 normalized compress_ratios (AMD + NVIDIA)#44031

dparikh79 wants to merge 1 commit into

vllm-project:mainvllm-project/vllm:mainfrom

dparikh79:fix/42741-deepseek-v4-models-compress-ratiosdparikh79/vllm:fix/42741-deepseek-v4-models-compress-ratiosCopy head branch name to clipboard