Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add support for IBM Granite 3.x models #2437

Merged
merged 7 commits into from
Dec 11, 2024
Merged

Conversation

frreiss
Copy link
Contributor

@frreiss frreiss commented Dec 11, 2024

Motivation

This PR adds support for inference on non-MoE IBM Granite 3.x models (see https://huggingface.co/collections/ibm-granite/granite-30-language-models-66fdb59bbb54785c3512114f).

Granite 3.0's architecture is similar to that of Llama 2 and 3, but with additional multiplier parameters for improved numerical stability through Maximal Update Parameterization.

Modifications

  • Added new model type for Granite models
  • Backported change to LogitsProcessor from vLLM 0.4. This change add an optional logits multiplier parameter. Granite models use this multiplier to implement Maximal Update Parameterization.
  • Added chat template for granite-3-instruct family
  • Added 2b parameter Granite model to regression suite

Checklist

  • Format your code according to the Contributor Guide.
  • Add unit tests as outlined in the Contributor Guide.
  • Update documentation as needed, including docstrings or example tutorials.

@merrymercy merrymercy merged commit 993956c into sgl-project:main Dec 11, 2024
15 checks passed
@merrymercy
Copy link
Contributor

Thanks for the contribution. It is merged!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants