Zero-shot Classification on EleutherAI (test)
64.7AccuracyWanda + SparseSwaps
Evaluation Results
| Method | Links | |
|---|---|---|
| Wanda + SparseSwapsModel=GEMMA-2 9B, Sparsity=2:42025.12 | 64.7 | |
| RIA + SparseSwapsModel=GEMMA-2 9B, Sparsity=2:42025.12 | 64.6 | |
| RIA + DSnoTModel=GEMMA-2 9B, Sparsity=60%2025.12 | 64.43 | |
| RIAModel=GEMMA-2 9B, Sparsity=60%2025.12 | 64.37 | |
| RIA + SparseSwapsModel=GEMMA-2 9B, Sparsity=60%2025.12 | 64.32 | |
| RIA + DSnoTModel=GEMMA-2 9B, Sparsity=2:42025.12 | 64.17 | |
| RIAModel=GEMMA-2 9B, Sparsity=2:42025.12 | 63.87 | |
| Wanda + SparseSwapsModel=GEMMA-2 9B, Sparsity=60%2025.12 | 63.84 | |
| WandaModel=GEMMA-2 9B, Sparsity=2:42025.12 | 63.73 | |
| Wanda + DSnoTModel=GEMMA-2 9B, Sparsity=2:42025.12 | 63.66 | |
| Wanda + DSnoTModel=GEMMA-2 9B, Sparsity=60%2025.12 | 63.49 | |
| WandaModel=GEMMA-2 9B, Sparsity=60%2025.12 | 63.39 | |
| RIA + SparseSwapsModel=QWEN2.5 7B, Sparsity=60%2025.12 | 61.22 | |
| RIA + SparseSwapsModel=QWEN2.5 7B, Sparsity=2:42025.12 | 60.31 | |
| Wanda + SparseSwapsModel=QWEN2.5 7B, Sparsity=60%2025.12 | 60.15 | |
| Wanda + SparseSwapsModel=QWEN2.5 7B, Sparsity=2:42025.12 | 59.92 | |
| RIAModel=QWEN2.5 7B, Sparsity=60%2025.12 | 59.84 | |
| RIA + DSnoTModel=QWEN2.5 7B, Sparsity=60%2025.12 | 59.81 | |
| RIA + DSnoTModel=QWEN2.5 7B, Sparsity=2:42025.12 | 59.72 | |
| WandaModel=QWEN2.5 7B, Sparsity=2:42025.12 | 59.52 | |
| WandaModel=QWEN2.5 7B, Sparsity=60%2025.12 | 59.26 | |
| Wanda + DSnoTModel=QWEN2.5 7B, Sparsity=60%2025.12 | 59.26 | |
| Wanda + DSnoTModel=QWEN2.5 7B, Sparsity=2:42025.12 | 59.09 | |
| RIAModel=QWEN2.5 7B, Sparsity=2:42025.12 | 58.66 | |
| Wanda + SparseSwapsModel=YI-1.5 9B, Sparsity=60%2025.12 | 54.84 | |
| RIA + SparseSwapsModel=YI-1.5 9B, Sparsity=60%2025.12 | 54.45 | |
| Wanda + DSnoTModel=YI-1.5 9B, Sparsity=60%2025.12 | 53.79 | |
| WandaModel=YI-1.5 9B, Sparsity=60%2025.12 | 53.59 | |
| RIA + DSnoTModel=YI-1.5 9B, Sparsity=60%2025.12 | 52.96 | |
| RIAModel=YI-1.5 9B, Sparsity=60%2025.12 | 52.81 | |
| RIAModel=YI-1.5 9B, Sparsity=2:42025.12 | 52.68 | |
| WandaModel=YI-1.5 9B, Sparsity=2:42025.12 | 52.58 | |
| Wanda + SparseSwapsModel=YI-1.5 9B, Sparsity=2:42025.12 | 52.43 | |
| RIA + SparseSwapsModel=YI-1.5 9B, Sparsity=2:42025.12 | 52.3 | |
| Wanda + DSnoTModel=YI-1.5 9B, Sparsity=2:42025.12 | 52.16 | |
| RIA + SparseSwapsModel=DEEPSEEK 7B, Sparsity=60%2025.12 | 51.47 | |
| RIA + SparseSwapsModel=DEEPSEEK 7B, Sparsity=2:42025.12 | 51.46 | |
| RIA + DSnoTModel=YI-1.5 9B, Sparsity=2:42025.12 | 51.36 | |
| RIAModel=DEEPSEEK 7B, Sparsity=2:42025.12 | 51.22 | |
| Wanda + SparseSwapsModel=DEEPSEEK 7B, Sparsity=60%2025.12 | 51.02 | |
| RIA + SparseSwapsModel=LLAMA-3.1 8B, Sparsity=60%2025.12 | 51.02 | |
| WandaModel=DEEPSEEK 7B, Sparsity=2:42025.12 | 51.02 | |
| RIAModel=DEEPSEEK 7B, Sparsity=60%2025.12 | 50.92 | |
| RIA + DSnoTModel=DEEPSEEK 7B, Sparsity=60%2025.12 | 50.83 | |
| Wanda + SparseSwapsModel=LLAMA-3.1 8B, Sparsity=60%2025.12 | 50.78 | |
| Wanda + DSnoTModel=DEEPSEEK 7B, Sparsity=2:42025.12 | 50.78 | |
| Wanda + DSnoTModel=DEEPSEEK 7B, Sparsity=60%2025.12 | 50.75 | |
| WandaModel=DEEPSEEK 7B, Sparsity=60%2025.12 | 50.74 | |
| Wanda + SparseSwapsModel=DEEPSEEK 7B, Sparsity=2:42025.12 | 50.36 | |
| RIA + SparseSwapsModel=LLAMA-3.1 8B, Sparsity=2:42025.12 | 49.9 | |
| RIA + DSnoTModel=DEEPSEEK 7B, Sparsity=2:42025.12 | 49.86 | |
| RIAModel=LLAMA-3.1 8B, Sparsity=60%2025.12 | 49.56 | |
| RIA + DSnoTModel=LLAMA-3.1 8B, Sparsity=60%2025.12 | 49.56 | |
| Wanda + SparseSwapsModel=LLAMA-3.1 8B, Sparsity=2:42025.12 | 48.83 | |
| WandaModel=LLAMA-3.1 8B, Sparsity=60%2025.12 | 48.18 | |
| Wanda + DSnoTModel=LLAMA-3.1 8B, Sparsity=60%2025.12 | 48.18 | |
| RIAModel=LLAMA-3.1 8B, Sparsity=2:42025.12 | 47.87 | |
| RIA + DSnoTModel=LLAMA-3.1 8B, Sparsity=2:42025.12 | 47.13 | |
| Wanda + DSnoTModel=LLAMA-3.1 8B, Sparsity=2:42025.12 | 47.01 | |
| WandaModel=LLAMA-3.1 8B, Sparsity=2:42025.12 | 46.8 |