Natural Language Inference on RTE (0-shot and 32-shot settings)
84.8Accuracy (0-shot)OPT-IML 175B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OPT-IML 175BScale=175B2022.12 | 84.8 | 83.8 | |
| OPT-IML 30BScale=30B2022.12 | 83.8 | 73.3 | |
| OPT-IML 1.3BScale=1.3B2022.12 | 66.8 | 48.7 | |
| DenseSparsity=0%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 65.34 | — | |
| SparseGPTSparsity=50%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 64.26 | — | |
| L-ADMMSparsity=50%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 63.18 | — | |
| ALPSSparsity=50%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 62.82 | — | |
| WandaSparsity=50%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 60.65 | — | |
| SAFESparsity=50%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 60.65 | — | |
| OPT 175BScale=175B2022.12 | 60.3 | 71 | |
| ALPSSparsity=60%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 60.29 | — | |
| L-ADMMSparsity=60%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 59.93 | — | |
| WandaSparsity=60%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 59.57 | — | |
| OPT 30BScale=30B2022.12 | 58.1 | 61.7 | |
| SparseGPTSparsity=60%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 57.76 | — | |
| SAFESparsity=60%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 57.4 | — | |
| MagnitudeSparsity=50%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 55.96 | — | |
| ELSASparsity=50%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 55.6 | — | |
| OPT 1.3BScale=1.3B2022.12 | 54.2 | 47.3 | |
| L-ADMMSparsity=90%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 53.79 | — | |
| ELSASparsity=60%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 53.43 | — | |
| L-ADMMSparsity=70%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 53.43 | — | |
| ALPSSparsity=70%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 53.07 | — | |
| MagnitudeSparsity=80%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 53.07 | — | |
| MagnitudeSparsity=60%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| MagnitudeSparsity=70%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| WandaSparsity=70%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| SparseGPTSparsity=70%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| SAFESparsity=70%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| ELSASparsity=70%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| WandaSparsity=80%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| SparseGPTSparsity=80%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| L-ADMMSparsity=80%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| ALPSSparsity=80%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| SAFESparsity=80%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| ELSASparsity=80%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| WandaSparsity=90%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| ALPSSparsity=90%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| SAFESparsity=90%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| ELSASparsity=90%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.71 | — | |
| SparseGPTSparsity=90%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 52.35 | — | |
| MagnitudeSparsity=90%, Backbone=Llama-2 13B, Protocol=Zero-shot2025.10 | 46.57 | — |