Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

VLLM assert isinstance(self.device, list) error #27

Closed
WanBenLe opened this issue May 12, 2024 · 2 comments
Closed

VLLM assert isinstance(self.device, list) error #27

WanBenLe opened this issue May 12, 2024 · 2 comments

Comments

@WanBenLe
Copy link

from datadreamer.llms import ParallelLLM,VLLM,LLM,HFTransformers
VLLM(Mistral-7B-Instruct-v0.2", device=[1]).model
will rasie device error
File ~/autodl-tmp/conda/envs/wslconda/lib/python3.10/site-packages/datadreamer/llms/vllm.py:74, in VLLM.model(self)
72 env = os.environ.copy()
73 print(self.device)
---> 74 assert isinstance(self.device, list)
75 env.update(get_device_env_variables(self.device))
76 kwargs = self.kwargs.copy()

@WanBenLe
Copy link
Author

and...
Process SpawnProcess-1:
Traceback (most recent call last):
File "/root/autodl-tmp/conda/envs/wslconda/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap
self.run()
File "/root/autodl-tmp/conda/envs/wslconda/lib/python3.10/multiprocessing/process.py", line 108, in run
self._target(*self._args, **self._kwargs)
File "/root/autodl-tmp/conda/envs/wslconda/lib/python3.10/site-packages/datadreamer/utils/background_utils.py", line 177, in _process_func_wrapper
restore_parent_process_context(parent_context=parent_context, env=env)
File "/root/autodl-tmp/conda/envs/wslconda/lib/python3.10/site-packages/datadreamer/utils/background_utils.py", line 155, in restore_parent_process_context
ctx.instance.enter()

AttributeError: 'UserDict' object has no attribute 'instance'

@WanBenLe
Copy link
Author

This package does not seem to implement parallel reasoning at all? According to my observation, it actually just repeats the same prompts to each model instead of sharding the prompts to the model, so there is no speed bonus for parallel reasoning at all.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant