Natural Language Inference on CrossFit NLI (test)
83.6AccuracyABMLL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ABMLLModel=QWEN22025.08 | 83.6 | 20.8 | |
| ReptileModel=LLAMA32025.08 | 83.3 | 24.2 | |
| ReptileModel=QWEN22025.08 | 82.6 | 23.1 | |
| ABMLLModel=LLAMA32025.08 | 82.2 | 23.7 | |
| Regular LoRAModel=QWEN22025.08 | 79.7 | 26.9 | |
| Struct. LoRAModel=QWEN22025.08 | 79.1 | 27 | |
| Regular LoRAModel=LLAMA32025.08 | 78.5 | 31 | |
| Struct. LoRAModel=LLAMA32025.08 | 75.5 | 30.2 | |
| PretrainedModel=QWEN22025.08 | 69.3 | 29.8 | |
| PretrainedModel=LLAMA32025.08 | 57.6 | 41.9 |