Long-context understanding on L-Eval
70.6Coursera AccuracyLPES
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LPESModel=Vicuna-13B-v1.5-16k, Context Length=16k2026.06 | 70.6 | 54.4 | 36 | 59.4 | 55.1 | |
| BaselineModel=Vicuna-13B-v1.5-16k, Context Length=16k2026.06 | 69.6 | 51.4 | 33.3 | 57.1 | 52.9 | |
| MoICEModel=Vicuna-13B-v1.5-16k, Context Length=16k2026.06 | 67.4 | 55.6 | 35.7 | 52.6 | 52.8 | |
| LPESModel=Qwen2.5-7B, Context Length=16k2026.06 | 63.8 | 69.1 | 80.9 | 72.9 | 71.7 | |
| BaselineModel=Qwen2.5-7B, Context Length=16k2026.06 | 59.8 | 66.3 | 76.6 | 71.8 | 68.6 | |
| MoICEModel=Qwen2.5-7B, Context Length=16k2026.06 | 59.8 | 66.3 | 78.7 | 73 | 69.5 |