Model request: a dedicated coding model

mlc-ai / web-llm

High-performance In-browser LLM Inference Engine

https://webllm.mlc.ai

Apache License 2.0

13.37k stars 857 forks source link

Model request: a dedicated coding model #449

Closed flatsiedatsie closed 3 weeks ago

flatsiedatsie commented 4 months ago

I could have sworn I saw CodeLlama 7B in WebLLM at some point, but it must have been a hallucination.

I was wondering if WebLLM could add support for a dedicated code instruct model. Something around 7B perhaps?

It seems there are discussions on Reddit about which coding model is good about every single week. The latest one being this one (May 2024).

CharlieFRuan commented 3 weeks ago

As you have probably noted, Qwen2.5-Coder-1.5B and Qwen2.5-Coder-7B are added to the prebuilt list as of ^0.2.63. Will close this issue for now, feel free to open new ones for other requests!