lm-sys / lm-sys/FastChat

fastchat.serve.model_worker --device cpu only uses one CPU Thread for token generation.

Open
#1,626 3 comments 2 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
39.5k
Forks
4.8k
PR merge metrics
No merged PRs in 30d

Description

Hi,

i launch a worker with python3 -m fastchat.serve.model_worker --model-path /home/llamaweights/vicuna-13b --device cpu and then the webGUI, which works fine so far. When i do a request, after an initial loading time one core goes to 100% while the others idle. If i make a second request in another tap another core goes to 100% while the other 14 idle... Token generation is very slow, but does not get any slower for additional requests. Can i somehow use all 16 threads or at least all 8 cores for a single request to speed up token generation?

Kind regards

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reproducing the behavior with python3 -m fastchat.serve.model_worker --model-path /home/llamaweights/vicuna-13b --device cpu and inspect the CPU execution path used by fastchat.serve.model_worker. Done means determining whether one request can use the available CPU threads and documenting or implementing a supported way to achieve that, if applicable.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
backend, machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.