Hallucination Evaluation on FINER-CompreCap 1.0 (whole)
80Multi-obj Acc (Paired)InternVL-3.5 + FINER-Tuning
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| InternVL-3.5 + FINER-TuningSize=14B, tuning=FINER-Tuning2026.03 | 80 | 78.9 | 71.2 | 30.1 | |
| InternVL-3.5Size=38B2026.03 | 77.8 | 78.1 | 66.8 | 50.9 | |
| InternVL-3.5 + FINER-TuningSize=8B, tuning=FINER-Tuning2026.03 | 77.1 | 78.9 | 64.1 | 34.2 | |
| Gemini-2.5-FlashSize=-2026.03 | 75.7 | 77.3 | 77.8 | 58.2 | |
| InternVL-3.5Size=8B2026.03 | 75 | 72.5 | 49.8 | 23.5 | |
| InternVL-3.5Size=14B2026.03 | 74.5 | 68.1 | 47 | 21.8 | |
| Qwen2.5-VL + FINER-TuningSize=7B, tuning=FINER-Tuning2026.03 | 71.4 | 67 | 38.3 | 34.8 | |
| Qwen2.5-VLSize=7B2026.03 | 69.2 | 62.5 | 30.1 | 28.9 | |
| RLAIF-VSize=12B2026.03 | 62.2 | 39.6 | 19.2 | 20.5 | |
| LLaVA-1.6 + FINER-TuningSize=7B, tuning=FINER-Tuning2026.03 | 48.4 | 38.4 | 24.2 | 22.1 | |
| LLaVA-1.6Size=7B2026.03 | 25.3 | 13 | 7.6 | 15.3 | |
| RLHF-VSize=13B2026.03 | 13.4 | 6.1 | 1.6 | 10.8 | |
| LLaVA-RLHFSize=13B2026.03 | 11.4 | 2 | 1.1 | 6.9 | |
| OPA-DPOSize=13B2026.03 | 10.9 | 3 | 2.2 | 6.9 | |
| LRV-V2Size=13B2026.03 | 6.1 | 6.8 | 5.6 | 4 | |
| Random GuessSize=-2026.03 | 4 | 4 | 4 | 4 |