General Reasoning on BigBench-Lite Topic Domains
66.8BBL ScoreQwen-3B (KOFF)
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen-3B (KOFF)Model Backbone=Qwen-3B, Model Variant=KOFF, Global Sparsity Level=12%2026.05 | 66.8 | |
| Llama-8B (KOFF)Model Backbone=Llama-8B, Model Variant=KOFF, Global Sparsity Level=12%2026.05 | 48 | |
| Llama-3B (KOFF)Model Backbone=Llama-3B, Model Variant=KOFF, Global Sparsity Level=12%2026.05 | 47 | |
| Llama-8B (unpruned)Model Backbone=Llama-8B, Model Variant=unpruned, Global Sparsity Level=0%2026.05 | 43.5 | |
| Qwen-3B (unpruned)Model Backbone=Qwen-3B, Model Variant=unpruned, Global Sparsity Level=0%2026.05 | 41 | |
| Llama-3B (unpruned)Model Backbone=Llama-3B, Model Variant=unpruned, Global Sparsity Level=0%2026.05 | 39.3 | |
| Llama-3B (pruning only)Model Backbone=Llama-3B, Model Variant=pruning only, Global Sparsity Level=12%2026.05 | 37.2 | |
| Qwen-3B (pruning only)Model Backbone=Qwen-3B, Model Variant=pruning only, Global Sparsity Level=12%2026.05 | 36 | |
| Llama-8B (pruning only)Model Backbone=Llama-8B, Model Variant=pruning only, Global Sparsity Level=12%2026.05 | 35.3 | |
| Llama-8B (unpruned)Model Backbone=Llama-8B, Model Variant=unpruned, Global Sparsity Level=0%2026.05 | 0.492 | |
| Llama-3B (KOFF)Model Backbone=Llama-3B, Model Variant=KOFF, Global Sparsity Level=12%2026.05 | 0.412 | |
| Qwen-3B (unpruned)Model Backbone=Qwen-3B, Model Variant=unpruned, Global Sparsity Level=0%2026.05 | 0.41 | |
| Llama-3B (unpruned)Model Backbone=Llama-3B, Model Variant=unpruned, Global Sparsity Level=0%2026.05 | 0.393 | |
| Llama-8B (KOFF)Model Backbone=Llama-8B, Model Variant=KOFF, Global Sparsity Level=12%2026.05 | 0.386 | |
| Llama-3B (pruning only)Model Backbone=Llama-3B, Model Variant=pruning only, Global Sparsity Level=12%2026.05 | 0.382 | |
| Qwen-3B (KOFF)Model Backbone=Qwen-3B, Model Variant=KOFF, Global Sparsity Level=12%2026.05 | 0.38 | |
| Qwen-3B (pruning only)Model Backbone=Qwen-3B, Model Variant=pruning only, Global Sparsity Level=12%2026.05 | 0.374 | |
| Llama-8B (pruning only)Model Backbone=Llama-8B, Model Variant=pruning only, Global Sparsity Level=12%2026.05 | 0.364 |