Eval bug: EAGLE3 with Qwen3.6 (qwen3_5 hybrid) target — missing t_layer_inp hooks, and llama_decode(ctx_dft) rc=-1 once context exceeds ~700

用户使用 llama-server 或 llama.cpp 的 EAGLE3 投机解码功能,目标模型为 unsloth/Qwen3.6-27B-MTP-GGUF (架构 qwen3_5 ,GDN hybrid),草稿模型为基于 Ex0bit/Qwen3.6-27B-PRISM-EAGLE3 转换的


![[Bug]: MTP Speculative Decoding with NVFP4: Weight Shape Mismatch](https://www.chat-gpts.plus/wp-content/uploads/2026/06/35031-2c47d432-768x403.jpg)



![[Bug]: ExcelParser.__call__ drops cells with value 0, False, or empty string](https://www.chat-gpts.plus/wp-content/uploads/2026/06/16313-cea78a57-768x403.jpg)
![[Bug]: /v1/chat/completions hangs when max_tokens is omitted (macOS Apple Silicon, 0.30.7 + 0.30.8)](https://www.chat-gpts.plus/wp-content/uploads/2026/06/16707-111391ad-768x403.jpg)
![[Question]: agent with web-crawler](https://www.chat-gpts.plus/wp-content/uploads/2026/06/5840-01201172-768x403.jpg)