Long-context Understanding on LongBench HotpotQA
52.15F1 ScoreBase+ctx
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Base+ctxLLM Backbone=Qwen3-4B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 52.15 | 51.78 | |
| Base+ctxLLM Backbone=Gemma-2-2B-It, Evaluation Protocol=Zero-shot2026.06 | 41 | 40.78 | |
| Doc2AtomLLM Backbone=Gemma-2-2B-It, Evaluation Protocol=Zero-shot2026.06 | 27.96 | 27.82 | |
| D2LckptLLM Backbone=Gemma-2-2B-It, Evaluation Protocol=Zero-shot2026.06 | 26.79 | 26.37 | |
| D2LrawLLM Backbone=Qwen3-4B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 25.88 | 25.88 | |
| D2LatomLLM Backbone=Qwen3-4B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 25.43 | 25.24 | |
| Doc2AtomLLM Backbone=Qwen3-4B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 24.77 | 24.67 | |
| D2LrawLLM Backbone=Gemma-2-2B-It, Evaluation Protocol=Zero-shot2026.06 | 22.3 | 22.21 | |
| D2LatomLLM Backbone=Gemma-2-2B-It, Evaluation Protocol=Zero-shot2026.06 | 19.49 | 19.49 | |
| D2LckptLLM Backbone=Qwen3-4B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 11.04 | 10.5 |