Whittle MoE 27B：从 Qwen3.8 切分出的路由器修复型混合专家模型
原标题：logic65/Qwen3.8-Whittle-MoE-27B-A17.8B
AI 摘要
Hugging Face 上发布了新模型 logic65/Qwen3.8-Whittle-MoE-27B-A17.8B，这是一个从 Qwen3.8-27B 中通过后验方式切分出的混合专家模型，总参数 27B，激活参数 17.8B。该模型通过仅训练路由器（冻结专家）恢复了性能，并新增了停止信号以解决推理中止问题。v2.2.1 修复了推理中止缺陷，但 GGUF 文件尚未更新。模型为研究预览版，计算预算已耗尽，依赖捐赠支持。
正文节选
--- license: apache-2.0 base_model: Qwen/Qwen3.8-27B tags: [moe, mixture-of-experts, research-preview, whittle, qwen3.5, gguf, router-healing] --- > ## v2.2.1 fixes the reasoning abort (29 Aug 2026) > > **Fixed in the safetensors weights at the root of this repo.** Served with > reasoning enabled, v2.2 could end its turn inside its own chain of thought > instead of answering. The stop gate had been trained on sequences where the chat > template opened a `<think>` block that the training data ne