[PR #14376] [MERGED] mlx: don't default to affine quantization for unquantized models #45891

Closed
opened 2026-04-25 01:30:07 -05:00 by GiteaMirror · 0 comments
Owner

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/14376
Author: @jessegross
Created: 2/23/2026
Status: Merged
Merged: 2/23/2026
Merged by: @jessegross

Base: mainHead: jessegross/mlx-quant


📝 Commits (1)

  • 9e48899 mlx: don't default to affine quantization for unquantized models

📊 Changes

2 files changed (+6 additions, -2 deletions)

View changed files

📝 x/imagegen/safetensors/loader.go (+3 -1)
📝 x/mlxrunner/model/quant.go (+3 -1)

📄 Description

Otherwise the BF16 version of models trigger segfaults when they call into quantized kernels.


🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/ollama/ollama/pull/14376 **Author:** [@jessegross](https://github.com/jessegross) **Created:** 2/23/2026 **Status:** ✅ Merged **Merged:** 2/23/2026 **Merged by:** [@jessegross](https://github.com/jessegross) **Base:** `main` ← **Head:** `jessegross/mlx-quant` --- ### 📝 Commits (1) - [`9e48899`](https://github.com/ollama/ollama/commit/9e48899dbeeb5563a71b9a08722d30829a886d10) mlx: don't default to affine quantization for unquantized models ### 📊 Changes **2 files changed** (+6 additions, -2 deletions) <details> <summary>View changed files</summary> 📝 `x/imagegen/safetensors/loader.go` (+3 -1) 📝 `x/mlxrunner/model/quant.go` (+3 -1) </details> ### 📄 Description Otherwise the BF16 version of models trigger segfaults when they call into quantized kernels. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the pull-request label 2026-04-25 01:30:07 -05:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#45891