LongVA-7B-DPO: benchmark results
Provider: Other.
Aggregate Rating 1522 ± 42, rank #2043 of 10916 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EgoIntrospect - Memory Recall Prediction | 49 | Accuracy (%) on 200 four-choice questions asking which detai | 72.5 |
| LongEgoRefer - Temporal Grounding | 1.3 | Mean temporal IoU (%; predicted versus annotated time interv | 50 |
| VRBench | 52.36 | Overall (imported) | 42.9 |
| AffectSim (Monitor View) | 8.95 | Macro-F1 (%; P-Mon: a fixed room-level monitoring camera; fi | 39.1 |
| AffectSim (Reference View) | 13.57 | Macro-F1 (%; P-Ref: an informative camera trajectory chosen | 39.1 |
| AffectSim (Active Observation) | 10.8 | Macro-F1 (%; A-Obs: the view reached by the authors' two-sta | 34.8 |
| AffectSim (Initial View) | 7.55 | Macro-F1 (%; P-Init: the agent's initial camera pose, held f | 26.1 |
| EgoIntrospect - Proactive Timing Judgment | 38.45 | Macro-F1 (%) over the interruptible and do-not-disturb class | 15 |
| EmoBench-M | 43.8 | Average Score | 4 |
| EgoIntrospect - Valuable Interaction Identification | 36.09 | Accuracy (%) on 1,161 four-choice questions asking which of | 0 |
Interactive version: theaggregate.ai/model?slug=longva-7b-dpo · How It Works · Data refreshed daily, snapshot 2026-11-08.