在 Mac 上基于 Apple Silicon 用 MLX 跑视觉语言模型的推理与微调工具包，支持 LLaVA、Pixtral、Florence2 等多种 VLM，本地即可完成多模态任务。
把屏幕截图解析成结构化元素，让纯视觉 GUI Agent 能识别界面上的按钮、图标和文本，无需依赖 DOM 或控件树。
…b.com/roboflow/inference/pull/3177). - Fine-tuned Florence-2-large answers again — LoRA fine-tunes returned empty strings because decoding repeated <s until the token lim
…[Ovis2](https://huggingface.co/AIDC-AI/Ovis2-2B) [Florence-2](https://huggingface.co/microsoft/Florence-2-base) [GLM-4.1V-9B-Thinking](https://huggingface.co/zai-org/GLM-
…ps://github.com/ultralytics/notebooks/pull/38 Add Florence-2 notebook by @RizwanMunawar in https://github.com/ultralytics/notebooks/pull/39 Fix florence-2 notebook colab
… it tough to keep up. Maestro currently supports [Florence-2](https://maestro.roboflow.com/latest/models/florence 2/), [PaliGemma 2](https://maestro.roboflow.com/latest/m