Change8

v0.20.4-rc1

📦 ollamaView on GitHub →
2 features🐛 2 fixes🔧 3 symbols

Summary

This release focuses on performance improvements for MLX (M5 with NAX) and Gemma4 (flash attention), alongside fixes for model creation paths and safetensor loading.

✨ New Features

  • Improve M5 performance with NAX for MLX backend.
  • Enable flash attention for Gemma4 models.

🐛 Bug Fixes

  • Clean up experimental paths during model creation.
  • Fix model creation when starting from an existing safetensor model.

Affected Symbols