SLVR:结构化潜在视觉推理框架
Get the story
研究者提出结构化潜在视觉推理框架 SLVR,将多模态推理组织为规划、定位、证据选择与整合的带类型潜在阶段,并以对应信号分阶段监督,在推理时避免生成文本链式推理。该框架基于 Qwen2.5-VL-7B 训练,在 MMVP 上提升 +9.4、BLINK Relation 上提升 +14.2,并在 V*、MathVista、ChartQA 上均有改进。论文已被 NeurIPS 接收。
Generated from reports · updated 3 hr ago
Timeline
Follow the coverage from different angles.
- arXiv · Computer VisionSLVR:基于类人推理流的结构化潜在视觉推理
研究者提出结构化潜在视觉推理框架 SLVR,将多模态推理组织为规划、定位、证据选择与整合的带类型潜在阶段,并以对应信号分阶段监督,在推理时避免生成文本链式推理。基于 Qwen2.5-VL-7B 训练,SLVR 在 MMVP 上提升 +9.4、BLINK Relation 上提升 +14.2,并在 V*、MathVista、ChartQA 上均有改进。论文已被 NeurIPS 接收。
Heat trend
Current heat 9·Comparable peak 10(Oct 9)·Comparable change over 24 hours –
The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.