Aggregate Downstream Performance on Aggregate Downstream Performance Suite
64.16Average AccuracyDense
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DenseBase Model=Llama-3.1-8B-Instruct, Sparsity=Dense2026.06 | 64.16 | 100 | |
| Llama-3.1-8B-InstructModel=Llama-3.1-8B-Instruct, Sparsity=Dense2026.06 | 64.16 | 100 | |
| DenseBase Model=DeepSeek-7B-chat, Sparsity=Dense2026.06 | 58.38 | 100 | |
| DeepSeek-7B-chatModel=DeepSeek-7B-chat, Sparsity=Dense2026.06 | 58.38 | 100 | |
| DenseBase Model=Llama-3.2-3B-Instruct, Sparsity=Dense2026.06 | 57.87 | 100 | |
| Llama-3.2-3B-InstructModel=Llama-3.2-3B-Instruct, Sparsity=Dense2026.06 | 57.87 | 100 | |
| SUBFITModel=Llama-3.1-8B-Instruct, Sparsity=25%2026.06 | 57.02 | 88.87 | |
| DenseBase Model=Qwen2.5-7B-Instruct, Sparsity=Dense2026.06 | 56.69 | 100 | |
| Qwen2.5-7B-InstructModel=Qwen2.5-7B-Instruct, Sparsity=Dense2026.06 | 56.69 | 100 | |
| DenseBase Model=Qwen3-4B-Instruct, Sparsity=Dense2026.06 | 54.38 | 100 | |
| Qwen3-4B-InstructModel=Qwen3-4B-Instruct, Sparsity=Dense2026.06 | 54.38 | 100 | |
| SUBFITModel=Llama-3.2-3B-Instruct, Sparsity=25%2026.06 | 51.36 | 88.75 | |
| SUBFITModel=DeepSeek-7B-chat, Sparsity=25%2026.06 | 49.15 | 84.19 | |
| SUBFITBase Model=Llama-3.1-8B-Instruct, Sparsity=37.5%2026.06 | 46.94 | 73.15 | |
| SUBFITModel=Qwen2.5-7B-Instruct, Sparsity=25%2026.06 | 45.66 | 80.54 | |
| SUBFITBase Model=Llama-3.2-3B-Instruct, Sparsity=37.5%2026.06 | 45.28 | 78.24 | |
| SUBFITModel=Qwen3-4B-Instruct, Sparsity=25%2026.06 | 43.76 | 80.48 | |
| SUBFITBase Model=Qwen2.5-7B-Instruct, Sparsity=37.5%2026.06 | 42.41 | 74.81 | |
| SUBFITBase Model=DeepSeek-7B-chat, Sparsity=37.5%2026.06 | 40.49 | 69.36 | |
| SUBFITBase Model=Qwen3-4B-Instruct, Sparsity=37.5%2026.06 | 40.31 | 74.13 |