Natural Language Inference on SciTail
96.52AccuracyIndividual Task
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Individual TaskBackbone=Llama-3, Evaluation Protocol=fine-tuned on single tasks2026.04 | 96.52 | — | |
| CoreSpace + ENMPBackbone=Llama-3, Scaling factor (λ)=0.5, ENMP (Elective Neuron Modification Protocol)=true2026.04 | 94.97 | — | |
| FinetunedBackbone=LLAMA-3 8B, LoRA rank=162026.03 | 94.8 | — | |
| CoreSpaceBackbone=Llama-3, Scaling factor (λ)=0.52026.04 | 94.45 | — | |
| TSV + ENMPBackbone=Llama-3, Scaling factor (λ)=0.6, ENMP (Elective Neuron Modification Protocol)=true2026.04 | 94.18 | — | |
| TSVBackbone=Llama-3, Scaling factor (λ)=0.62026.04 | 94.17 | — | |
| TA + ENMPBackbone=Llama-3, Scaling factor (λ)=0.3, ENMP (Elective Neuron Modification Protocol)=true2026.04 | 93.98 | — | |
| KnOTSBackbone=Llama-3, Scaling factor (λ)=1.1, Top-k (%)=902026.04 | 93.98 | — | |
| KnOTS + ENMPBackbone=Llama-3, Scaling factor (λ)=1.1, Top-k (%)=90, ENMP (Elective Neuron Modification Protocol)=true2026.04 | 93.85 | — | |
| DARE + ENMPBackbone=Llama-3, Scaling factor (λ)=1.1, Top-k (%)=80, Drop rate (p)=0.1, ENMP (Elective Neuron Modification Protocol)=true2026.04 | 93.84 | — | |
| TABackbone=Llama-3, Scaling factor (λ)=0.32026.04 | 93.32 | — | |
| TIES + ENMPBackbone=Llama-3, Scaling factor (λ)=1.2, Top-k (%)=80, ENMP (Elective Neuron Modification Protocol)=true2026.04 | 93.04 | — | |
| DAREBackbone=Llama-3, Scaling factor (λ)=1.1, Top-k (%)=80, Drop rate (p)=0.12026.04 | 92.8 | — | |
| TIESBackbone=Llama-3, Scaling factor (λ)=1.2, Top-k (%)=802026.04 | 92.66 | — | |
| DPS MixTraining Dataset=MNLI2022.11 | 80.36 | — | |
| CHILD-TUNINGDTraining Dataset=MNLI2022.11 | 79.86 | — | |
| vanillaTraining Dataset=MNLI2022.11 | 79.7 | — | |
| DPS DenseTraining Dataset=MNLI2022.11 | 79.35 | — | |
| DPS DenseTraining Dataset=SNLI2022.11 | 78.42 | — | |
| vanillaTraining Dataset=SNLI2022.11 | 78.22 | — | |
| DPS MixTraining Dataset=SNLI2022.11 | 77.8 | — | |
| CHILD-TUNINGDTraining Dataset=SNLI2022.11 | 77.17 | — | |
| MICBackbone=TinyBERT 6L, Dimension=7682026.05 | 74.5 | — | |
| MICBackbone=BERT, Dimension=7682026.05 | 74.18 | — | |
| MICBackbone=BERT, Dimension=162026.05 | 73.09 | — | |
| MICBackbone=TinyBERT 6L, Dimension=162026.05 | 71.78 | — | |
| AdaMergingBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 84.8 | |
| DARE-TIESBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 94.4 | |
| EMR-MergingBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 94.5 | |
| FR-MergingBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 92.6 | |
| KnOTS-DARE-TIESBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 95.4 | |
| KnOTS-TIESBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 95.3 | |
| LinearBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 96.1 | |
| LoRA-LEGOBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 95.6 | |
| NSCBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 95.1 | |
| RobustMergeBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 94.5 | |
| SVDBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 92.4 | |
| TABackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 95 | |
| TIESBackbone=LLAMA-3 8B, LoRA rank=162026.03 | — | 94.4 |