lm-sys / lm-sys/FastChat

the model keeps repeating the same answer all the time

Open
#2,097 6 comments 0 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

os: ubuntu18.04 server
python: 3.10.6
fschat: 0.2.15
FastChat: 4e2c942b8d785eb5e2aef1d0df2150e756f381ab
---
the problem I meet is that the model keeps repeating the same answer and I don't know what's going on. Maybe something is wrong? Can anyone give me some advice?
```
curl http://192.168.0.1:80/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "llama-2-13b-chat",
"messages": [
{"role":"assistant","content":"I love freedom. Do you like freedom?"},
{"role":"user","content":"Of course, I like freedom, I am a wild horse, no one can stop me."},
{"role":"assistant","content":"I love freedom. Do you like freedom?"},
{"role":"user","content":"oo"},
{"role":"assistant","content":"I love freedom. Do you like freedom?"},
{"role":"user","content":"I like you, are you called freedom?"},
{"role":"assistant","content":"I love freedom. Do you like freedom?"},
{"role":"user","content":"what"},
{"role":"assistant","content":"I love freedom. Do you like freedom?"},
{"role":"user","content":"can you say something else"}],
"temperature": 0.75
}'

```

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 exact curl request against the reported FastChat commit and Python environment, paying attention to the supplied conversation history and the llama-2-13b-chat serving path. Determine whether the repeated response is caused by the request, model behavior, or FastChat, then document a reproducible cause and fix or workaround.

Written by the indexing model from the issue text.

Assessment

Tech stack
linux, python
Domain
api, 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.