Question Answering on HOTPOTQA Unfiltered
66.08AccuracyBaseline
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| BaselineModel=Qwen 2.5 32B2026.05 | 66.08 | 73.95 | 2.95 | |
| +ScramblerModel=Qwen 2.5 32B2026.05 | 65.64 | 73.48 | 2.96 | |
| +DistributedModel=Qwen 2.5 32B2026.05 | 65.56 | 73.44 | 2.96 | |
| +DistributedModel=Mistral 7B2026.05 | 61.74 | 71.51 | 1.9 | |
| +ScramblerModel=Mistral 7B2026.05 | 61.66 | 71.39 | 1.89 | |
| BaselineModel=Mistral 7B2026.05 | 61.63 | 71.41 | 1.89 | |
| +DistributedModel=Qwen 2 7B2026.05 | 61.08 | 69.61 | 11.63 | |
| BaselineModel=Qwen 2 7B2026.05 | 61.01 | 69.59 | 11.62 | |
| +ScramblerModel=Qwen 2 7B2026.05 | 60.93 | 69.56 | 11.83 | |
| +DistributedModel=Llama 3.1 8B2026.05 | 56.96 | 66.02 | 2.15 | |
| BaselineModel=Llama 3.1 8B2026.05 | 56.87 | 66.01 | 2.14 | |
| +ScramblerModel=Llama 3.1 8B2026.05 | 56.87 | 65.92 | 2.15 |