lm-sys / lm-sys/FastChat

Possibility of hiding aliases from the WEB UI but allowing it on the API endpoints

Open
#2,957 1 comment 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

As in production, many models might come and go, it's difficult to have tools (e.g.: https://continue.dev ) which rely on hardcoded names for the models/workers it wants to use.

For that, the use of aliases makes sense. You can set, for example, an alias to "Mistral" regardless of the mistral version in use (right now it's 0.2).

I set, for example, aliases for "fast", "code" and "large", which at some point in time might point to the current iteration of mistral, starcoder and llama2-70, for example.

But by setting aliases, one also duplicate the entries at the list of available models on the web UI.

What do you think about a way to not show an aliased model? Or the possibility to exclude some models from the UI?

I have been thinking of checking for a string, say "alias-code" and models with this name are simply ignored by the UI. Another way would be to have a command line parameter telling which ones to ignore, or a config file.

What's your opinion?

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 tracing how model aliases are represented in the WEB UI and exposed through the API endpoints. The issue proposes hiding aliased or excluded models from the UI while keeping them available through the API, but leaves the configuration mechanism undecided; done would require an agreed behavior and an implementation that preserves API access.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
api, frontend
Issue type
Feature
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.