Research map / Video / Video-language understanding
Enhancing Video Vision Language Model with Hippocampal Sensing
Xu Cao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add VANAO RL post-training after HippoSense SFT (ablation) | HippoVLM-7B · Video-MME (w/o sub), 16 frames | accuracy: 59.1 -> 62.5 | Improved |
| add VANAO RL post-training after HippoSense SFT (ablation) | HippoVLM-7B · VNBench, 16 frames | accuracy: 57.3 -> 66.0 | Improved |
| add VANAO RL post-training after HippoSense SFT (ablation) | HippoVLM-7B · Social-IQ 2.0, 16 frames | accuracy: 63.4 -> 69.5 | Improved |
| add HippoSense cross-modal SFT instead of plain SFT (ablation) | Qwen2.5-VL-7B · Video-MME (w/o sub), 16 frames | accuracy: 53.3 -> 59.1 | Improved |
| add HippoSense cross-modal SFT instead of plain SFT (ablation) | Qwen2.5-VL-7B · VNBench, 16 frames | accuracy: 44.1 -> 57.3 | Improved |
| add HippoSense cross-modal SFT instead of plain SFT (ablation) | Qwen2.5-VL-7B · VideoMMMU, 16 frames | accuracy: 48.0 -> 48.3 | No clear effect |
| plain SFT without HippoSense (ablation) | Qwen2.5-VL-7B · Social-IQ 2.0, 16 frames | accuracy: 62.0 -> 60.2 | Hurt |
| HippoSense + VANAO instead of plain SFT+GRPO (ablation) | Qwen2.5-VL-7B · Video-MME (w/o sub), 16 frames | accuracy: 59.8 -> 62.5 | Improved |
| HippoSense + VANAO instead of plain SFT+GRPO (ablation) | Qwen2.5-VL-7B · VNBench, 16 frames | accuracy: 64.4 -> 66.0 | Mixed |
| increase frames from 16 to 64 | HippoVLM-7B · Video-MME (w/o sub) | accuracy: 62.5 -> 68.8 | Improved |
| two-stage summarize-then-reason pipeline | HippoVLM-7B · inference time, 16 frames | seconds: 7.231 -> 18.150 | Hurt |
Closest papers by meaning
- Bisecle: Binding and Separation in Continual Learning for Video Language Understanding (NeurIPS 2025 · similarity 0.88)
- Empowering Small VLMs to Think with Dynamic Memorization and Exploration (ICLR 2026 · similarity 0.86)
- Video-R1: Reinforcing Video Reasoning in MLLMs (NeurIPS 2025 · similarity 0.86)
- Beyond Perceptual Shortcuts: Causal-Inspired Debiasing Optimization for Generalizable Video Reasoning in Lightweight MLLMs (CVPR 2026 · similarity 0.85)
- TTRV: Test-Time Reinforcement Learning for Vision Language Models (CVPR 2026 · similarity 0.85)
- EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
