Skip to content

Conversation

ytzi
Copy link
Contributor

@ytzi ytzi commented May 20, 2025

Adding a vllm generation backend that should be compatible with the original generation backend using Transformers pipeline.

Important difference compared to original backend:

  • Supports multi-gpu;
  • no batch_size parameter: vllm dynamically chooses the most optimal batch size.

@Dando18 Dando18 self-requested a review June 10, 2025 11:07
@Dando18 Dando18 merged commit dbb9539 into parallelcodefoundry:develop Jun 10, 2025
61 checks passed
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.

2 participants