JiRack Ultra 1B:面向 CPU 的三值量化小模型
原标题:CMSManhattan/JiRackUltra_1b
AI 摘要
CMSManhattan 在 Hugging Face 发布 JiRack Ultra 1B 模型,约 1.5B 参数,基于重新设计的 DeepSeek R1 架构并原生支持 BitNet 风格三值(1.58bit)量化,面向 CPU 推理优化,提供 GGUF 量化版本与 Docker 部署方式。模型更新了 tokenizer,新增 Routing、Tool call、Robotics 标签,并配套 JiRack Coding Agent IDE 插件(Eclipse、JetBrains、VS Code)及 Ollama 支持。
正文节选
--- language: - en - zh - ja - ko - fr - es - pt - de - it - ru - ar - vi - th tags: - text-generation - ternary - bitnet - 1.58bit - cpu - gguf - qwen2.5 - deepseek - efficient - low-memory - jirack - web-ui - routing - tool-call - robotics license: mit --- # JiRack Ultra 1B (CPU) A fast and efficient ~1.5B model optimized for CPU inference. The model was refactored with BitNet features and an updated tokenizer that includes new **Routing**, **Tool call**, and **Robotics** tags. Built on a rede