GitLab

docker pull image.sourcefind.cn:5000/dcu/admin/base/custom:lmdeploy0.0.13_dtk23.04_torch1.13_py38

用上面拉取docker镜像的ID替换

主机端路径

容器映射路径

docker run -it --name llama --shm-size=1024G --device=/dev/kfd --device=/dev/dri/ --cap-add=SYS_PTRACE --security-opt seccomp=unconfined --ulimit memlock=-1:-1 --ipc=host --network host --group-add video -v : /bin/bash

running: 运行 LLama2-13b

模型转换

lmdeploy convert --model_name llama2 --model_path /path/to/model --model_format hf --tokenizer_path None --dst_path ./workspace_llama2-13b --tp 1

bash界面运行

lmdeploy chat turbomind --model_path ./workspace_llama2-13b --tp 1

error: [root@master141-k100 llama_lmdeploy]# lmdeploy chat turbomind --model_path ./workspace --tp 8 Traceback (most recent call last): File "/usr/local/bin/lmdeploy", line 33, in sys.exit(load_entry_point('lmdeploy==0.0.13', 'console_scripts', 'lmdeploy')()) File "/usr/local/lib/python3.8/site-packages/lmdeploy-0.0.13-py3.8-linux-x86_64.egg/lmdeploy/cli/cli.py", line 61, in run fire.Fire(cli, name='lmdeploy') File "/usr/local/lib/python3.8/site-packages/fire/core.py", line 141, in Fire component_trace = _Fire(component, args, parsed_flag_args, context, name) File "/usr/local/lib/python3.8/site-packages/fire/core.py", line 475, in _Fire component, remaining_args = _CallAndUpdateTrace( File "/usr/local/lib/python3.8/site-packages/fire/core.py", line 691, in _CallAndUpdateTrace component = fn(*varargs, **kwargs) File "/usr/local/lib/python3.8/site-packages/lmdeploy-0.0.13-py3.8-linux-x86_64.egg/lmdeploy/cli/chat.py", line 85, in turbomind run_turbomind_model(model_path, File "/usr/local/lib/python3.8/site-packages/lmdeploy-0.0.13-py3.8-linux-x86_64.egg/lmdeploy/turbomind/chat.py", line 92, in main tm_model = tm.TurboMind(model_path, eos_id=tokenizer.eos_token_id, tp=tp) File "/usr/local/lib/python3.8/site-packages/lmdeploy-0.0.13-py3.8-linux-x86_64.egg/lmdeploy/turbomind/turbomind.py", line 128, in init self.nccl_params = model.create_nccl_params(self.node_id) RuntimeError: [TM][ERROR] Assertion fail: /data/Imdeploy/data_2304/0.0.13/src/lmdeploy/src/turbomind/triton_backend/transformer_triton_backend.cpp:32

[root@master141-k100 llama_lmdeploy]# lmdeploy chat turbomind --model_path ./workspace --tp 8 Traceback (most recent call last): File "/usr/local/bin/lmdeploy", line 33, in sys.exit(load_entry_point('lmdeploy==0.0.13', 'console_scripts', 'lmdeploy')()) File "/usr/local/lib/python3.8/site-packages/lmdeploy-0.0.13-py3.8-linux-x86_64.egg/lmdeploy/cli/cli.py", line 61, in run fire.Fire(cli, name='lmdeploy') File "/usr/local/lib/python3.8/site-packages/fire/core.py", line 141, in Fire component_trace = _Fire(component, args, parsed_flag_args, context, name) File "/usr/local/lib/python3.8/site-packages/fire/core.py", line 475, in _Fire component, remaining_args = _CallAndUpdateTrace( File "/usr/local/lib/python3.8/site-packages/fire/core.py", line 691, in _CallAndUpdateTrace component = fn(*varargs, **kwargs) File "/usr/local/lib/python3.8/site-packages/lmdeploy-0.0.13-py3.8-linux-x86_64.egg/lmdeploy/cli/chat.py", line 85, in turbomind run_turbomind_model(model_path, File "/usr/local/lib/python3.8/site-packages/lmdeploy-0.0.13-py3.8-linux-x86_64.egg/lmdeploy/turbomind/chat.py", line 92, in main tm_model = tm.TurboMind(model_path, eos_id=tokenizer.eos_token_id, tp=tp) File "/usr/local/lib/python3.8/site-packages/lmdeploy-0.0.13-py3.8-linux-x86_64.egg/lmdeploy/turbomind/turbomind.py", line 128, in init self.nccl_params = model.create_nccl_params(self.node_id) RuntimeError: [TM][ERROR] Assertion fail: /data/Imdeploy/data_2304/0.0.13/src/lmdeploy/src/turbomind/triton_backend/transformer_triton_backend.cpp:32