00000512
get /SDK_4.4.0/x86/sdk/ corex-docker-installer-4.4.0-10.2-ubuntu20.04-llm-py3.10-x86_64.run /home
get /client_tmp/support/w4a8/* /home/scripts
bash /home/corex-docker-installer-4.4.0-10.2-ubuntu20.04-llm-py3.10-x86_64.run
全选安装
docker run -dit -v /usr/src:/usr/src -v /lib/modules:/lib/modules -v /dev:/dev -v /home:/home --network=host --name=test --pid=host --ipc=host --privileged --cap-add=ALL corex:4.4.0 /bin/bash
docker exec -it test bash
cd /home/scripts
python3 w4a8_fp82int8.py --input-fp8-hf-path /home/model/DeepSeek-R1/ --output-int8-hf-path /home/model/DeepSeek-R1-w4a8/ --group-size -1 --format TN --version 2 --split-count 1
export TORCHDYNAMO_DISABLE=1
export VLLM_W8A8_MOE_USE_W4A8=1
export VLLM_KV_DISABLE_CROSS_GROUP_SHARE=1
export VLLM_PP_LAYER_PARTITION="31,30"
export VLLM_MLA_DISABLE=0
export VLLM_ENFORCE_CUDA_GRAPH=1
python3 -m vllm.entrypoints.openai.api_server --model /data/model/DeepSeek-R1-w4a8/ --tensor-parallel-size 8 --pipeline-parallel-size 2 --trust-remote-code --max-model-len 8192 --host 127.0.0.1 --port 1234 --gpu_memory_utilization 0.90 --enable-prefix-caching --disable-log-requests