Misc. bug: llama-quantize fails on DeepSeek-V4 — i32 ffn_gate_tid2eid routing table not excluded from quantization

用户在 macOS/Metal 环境下,使用 llama-quantize 工具对 DeepSeek-V4(如 deepseek-ai/DeepSeek-V4-Flash-DSpark )的 GGUF 文件进行量化时触发。典型命令为: llama-quantize --allow-requantiz



![[Bug]: I have add Qwen3.5-27B model from model provider config, but not use from chat application](https://www.chat-gpts.plus/wp-content/uploads/2026/07/13985-a8bb0926-768x403.jpg)




![[Question]: why Entities extraction of chunk 0 1/1 done, 0 nodes, 0 edges](https://www.chat-gpts.plus/wp-content/uploads/2026/07/13411-4ebcf4c1-768x403.jpg)