Deception Detection on XCHECK 1.0 (test)
87.44F1 ScoreNVILA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| NVILA# Params=15B, Strategy=ACCNOTE2026.03 | 87.44 | 87.45 | 87.56 | 87.32 | |
| NVILA# Params=15B, Strategy=Closed-book2026.03 | 85.76 | 85.16 | 82.39 | 89.43 | |
| Qwen2.5-VL# Params=32B, Strategy=ACCNOTE2026.03 | 84.08 | 82.31 | 76.41 | 93.47 | |
| NVILA# Params=8B, Strategy=ACCNOTE2026.03 | 83.34 | 84.79 | 92.1 | 76.1 | |
| NVILA# Params=15B, Strategy=Naive RAG2026.03 | 81.04 | 78.17 | 71.63 | 93.29 | |
| Qwen2.5-VL# Params=7B, Strategy=ACCNOTE2026.03 | 80.47 | 77.16 | 70.28 | 94.12 | |
| Qwen2.5-VL# Params=32B, Strategy=Closed-book2026.03 | 79.31 | 76.29 | 70.34 | 90.9 | |
| Qwen2.5-VL# Params=32B, Strategy=Naive RAG2026.03 | 78.89 | 76.56 | 71.76 | 87.59 | |
| NVILA# Params=8B, Strategy=Closed-book2026.03 | 76.9 | 80.84 | 96.79 | 63.79 | |
| InternVL3# Params=14B, Strategy=ACCNOTE2026.03 | 75.63 | 71.6 | 66.23 | 88.14 | |
| NVILA# Params=8B, Strategy=Naive RAG2026.03 | 75.32 | 78.95 | 91.02 | 64.25 | |
| InternVL3# Params=14B, Strategy=Closed-book2026.03 | 74.51 | 70.4 | 65.44 | 86.49 | |
| InternVL3# Params=8B, Strategy=ACCNOTE2026.03 | 74.05 | 69.21 | 63.99 | 87.87 | |
| LLaVA-NeXT# Params=32B, Strategy=Closed-book2026.03 | 73.43 | 65.58 | 59.79 | 95.13 | |
| InternVL3# Params=8B, Strategy=Naive RAG2026.03 | 72.76 | 70.67 | 67.94 | 78.31 | |
| InternVL3# Params=14B, Strategy=Naive RAG2026.03 | 72.56 | 66.57 | 61.56 | 88.33 | |
| GPT5-mini# Params=-, Strategy=Web search2026.03 | 72.43 | 66.13 | 61.07 | 88.97 | |
| LLaVA-NeXT# Params=32B, Strategy=ACCNOTE2026.03 | 72.33 | 70.36 | 67.82 | 77.48 | |
| Qwen2.5-VL# Params=7B, Strategy=Closed-book2026.03 | 72.28 | 62.5 | 57.33 | 97.79 | |
| LLaVA-NeXT# Params=32B, Strategy=Naive RAG2026.03 | 72.15 | 63.91 | 58.75 | 93.47 | |
| GPT5-mini# Params=-, Strategy=Closed-book2026.03 | 71.6 | 63.14 | 58.24 | 92.92 | |
| InternVL3# Params=8B, Strategy=Closed-book2026.03 | 71.51 | 70.45 | 69.03 | 74.17 | |
| Qwen2.5-VL# Params=7B, Strategy=Naive RAG2026.03 | 70.25 | 58.27 | 54.58 | 98.53 | |
| LLaVA-NeXT# Params=8B, Strategy=Naive RAG2026.03 | 63.85 | 55.97 | 54.16 | 77.76 | |
| LLaVA-NeXT# Params=8B, Strategy=Closed-book2026.03 | 62.29 | 55.1 | 53.69 | 74.17 | |
| LLaVA-NeXT# Params=8B, Strategy=ACCNOTE2026.03 | 61.3 | 54.69 | 53.49 | 71.78 | |
| SNIFFER# Params=∼26B, Strategy=Closed-book2026.03 | 42.02 | 58.27 | 68.83 | 30.24 | |
| SNIFFER# Params=∼26B, Strategy=Web search2026.03 | 35.02 | 56.34 | 68.45 | 23.53 |