LAION-AI / LAION-AI/Open-Assistant

Race condition with "events" causes short answers to be completely lost.

Open
#2,738 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

inference website
Dominant language
Python
Stars
37.4k
Forks
3.3k
PR merge metrics
No merged PRs in 30d

Description

The api goes through a 3-step process: prompter_message, assistant_message, and "events". This ONLY works in low-latency situations. "Events" is entirely at the mercy of how long the LLM will take to generate the text. If user specifically asked for a 1-word answer, and you're using a high-latency browser, the answer will be lost forever because the "event" will be missed by the time you asked for it. To top it off, Open Assistant does not appear to follow the standard protocol for "Last-Event-ID", meaning you can't ask the server to re-send the missed event

Contributor guide

Open the contributing guide

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 tracing the API's three-step sequence: prompter_message, assistant_message, and events, then inspect how event delivery behaves when generation is very fast or the browser is high-latency. Check the handling of Last-Event-ID and define completion as preserving short answers and allowing missed events to be requested again according to the standard protocol.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
api, backend
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.