b9850
📦 llama-cppView on GitHub →
✨ 2 features🐛 2 fixes🔧 3 symbols
Summary
This release introduces model registration updates for Qwen3next, fixes input assignment bugs, and addresses DFLASH issues for qwen-coder-next.
✨ New Features
- Registered t_layer_inp for Qwen3next model.
- Added tensor for attention normalization in Qwen3 model.
🐛 Bug Fixes
- Fixed input assignment in layer processing loop.
- Fixed DFLASH for qwen-coder-next.