Skip to content

โšก VLM Efficiency

๐Ÿ’ฌ EMNLP2026 ยท 1 paper notes

๐Ÿ“Œ Same area in other venues: ๐Ÿง  NeurIPS2026 (10) ยท ๐ŸŽž๏ธ ECCV2026 (57) ยท ๐Ÿ“ท CVPR2026 (63) ยท ๐Ÿ”ฌ ICLR2026 (18) ยท ๐Ÿ’ฌ ACL2026 (6) ยท ๐Ÿงช ICML2026 (4)

Learning to Zoom Efficiently with a Contrastive Curriculum

Addressing the heavy reliance of multimodal large language models on expensive human annotations and trajectory-level warm-start supervised fine-tuning (SFT) for acquiring zoom-in tool capabilities, this work presents a self-supervised intrinsic exploration reward formulated as an InfoNCE-style difference over subsequent reasoning logits coupled with an IoU-guided negative bounding box curriculum, achieving Pareto-optimal accuracy on high-resolution VQA benchmarks with just a single tool call.