Question Answering on NQ (EM, F1)
59.46EMPrompt tuning
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Prompt tuningBackbone=Llama-3.2-3B-Base, Evaluation Protocol=Prompt tuning, Context Setting=w/ context2026.02 | 59.46 | 73.32 | |
| Prompt tuningBackbone=Llama-3.2-1B-Base, Evaluation Protocol=Prompt tuning, Context Setting=w/ context2026.02 | 53.03 | 66.59 | |
| ComprExITBackbone=Llama-3.2-3B-Base, Evaluation Protocol=Context Compression2026.02 | 52.23 | 68.95 | |
| 500xBackbone=Llama-3.2-3B-Base, Evaluation Protocol=Context Compression2026.02 | 51.32 | 66.6 | |
| ICAEBackbone=Llama-3.2-3B-Base, Evaluation Protocol=Context Compression2026.02 | 50.87 | 66.75 | |
| ComprExITBackbone=Llama-3.2-1B-Base, Evaluation Protocol=Context Compression2026.02 | 45.93 | 63.88 | |
| BeaconBackbone=Llama-3.2-3B-Base, Evaluation Protocol=Context Compression2026.02 | 45.05 | 59.56 | |
| ICAEBackbone=Llama-3.2-1B-Base, Evaluation Protocol=Context Compression2026.02 | 42.26 | 58 | |
| Zero-shotBackbone=Llama-3.2-3B-Base, Evaluation Protocol=Zero-shot, Context Setting=w/ context2026.02 | 32.74 | 52.09 | |
| BeaconBackbone=Llama-3.2-1B-Base, Evaluation Protocol=Context Compression2026.02 | 21.04 | 37.05 | |
| Zero-shotBackbone=Llama-3.2-1B-Base, Evaluation Protocol=Zero-shot, Context Setting=w/ context2026.02 | 18.14 | 38.9 | |
| Zero-shotBackbone=Llama-3.2-3B-Base, Evaluation Protocol=Zero-shot, Context Setting=w/o context2026.02 | 10.31 | 18.47 | |
| 500xBackbone=Llama-3.2-1B-Base, Evaluation Protocol=Context Compression2026.02 | 5.58 | 13.89 | |
| Zero-shotBackbone=Llama-3.2-1B-Base, Evaluation Protocol=Zero-shot, Context Setting=w/o context2026.02 | 3.37 | 8.43 |