search

microsoft / onnxruntime-genai

Generative AI extensions for onnxruntime
MIT License
259 stars 65 forks source link

Fix packed QKV Add after packed QKV MatMul in model builder #432

Closed kunal-vaishnavi closed 1 month ago

kunal-vaishnavi commented 1 month ago

Description

This PR adds support for packing the bias after a packed QKV MatMul.

Motivation and Context

This PR is a follow up to this PR.