Long-context Question Answering on Qasper (Extract/Abstract/Bool F1)
54.57Extract F1GLM-4.1V-9B-Thinking VERA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GLM-4.1V-9B-Thinking VERARAG Strategy=Attention-Guided RAG2026.02 | 54.57 | 34.67 | 94.23 | |
| GLM-4.1V-9B-Thinking ColPali RAGRAG Strategy=ColPali RAG2026.02 | 47.51 | 26.74 | 75.16 | |
| GLM-4.1V-9B-Thinking Random RAGRAG Strategy=Random RAG2026.02 | 46.92 | 29.61 | 80.47 | |
| GLM-4.1V-9B-Thinking Embedding RAGRAG Strategy=Embedding RAG2026.02 | 45.21 | 23.71 | 73.6 | |
| GLM-4.1V-9B-ThinkingRAG Strategy=Direct (None)2026.02 | 41.69 | 26.71 | 81.28 | |
| Qwen3-VL-8B-Instruct VERARAG Strategy=Attention-Guided RAG2026.02 | 39.29 | 22.56 | 71.12 | |
| GlyphRAG Strategy=Direct (None)2026.02 | 39.26 | 24.45 | 45.62 | |
| Qwen3-VL-8B-InstructRAG Strategy=Direct (None)2026.02 | 25.54 | 17.82 | 70.36 | |
| Qwen3-VL-8B-Instruct OCR RAGRAG Strategy=OCR RAG2026.02 | 25.15 | 17.99 | 71.48 | |
| Qwen3-VL-8B-Instruct Random RAGRAG Strategy=Random RAG2026.02 | 23.96 | 16.24 | 65.86 |