Qwen2.5-VL-3B

Researchers at CERIST in Algeria released a fine-tuned Qwen2.5-VL-3B model for Arabic historical manuscript recognition, achieving a 6.58% character error rate and 22.34% word error rate. The model, a…

A new arXiv preprint (2607.27566v1) finds that direct answer-only supervised fine-tuning (SFT) on MedGemma-1.5-4B is the most robust adaptation method for multi-frame medical visual question answering…

ChronoStitch, a training-free method for composing independently stored visual KV memories, improves long-video question answering by re-basing post-rotary keys onto a global three-axis multimodal RoP…