| { | |
| "timestamp": "2026-01-24T17:42:16.456628", | |
| "base_model": "Qwen/Qwen3-4B-Base", | |
| "code_model": "Montalte/code_think", | |
| "math_model": "Montalte/math_think", | |
| "save_path": "/work1/yizhanh/mhm/MergeBench/merged_models/unified_hf_code_math_qwen3_4b", | |
| "code_skill_region": { | |
| "actual_sparsity": 0.10032479662122679, | |
| "masked_energy_ratio": 0.9867451274256425, | |
| "dataset_size": 50 | |
| }, | |
| "math_skill_region": { | |
| "actual_sparsity": 0.10630752363104388, | |
| "masked_energy_ratio": 0.9861190180002117, | |
| "dataset_size": 50 | |
| }, | |
| "skill_regions_overlap": { | |
| "jaccard_similarity": 0.1623715126053168, | |
| "intersection_size": 104879288, | |
| "union_size": 645921728, | |
| "mask_1_size": 364531328, | |
| "mask_2_size": 386269664 | |
| }, | |
| "graft_args": { | |
| "sparsity": 0.1, | |
| "sigmoid_bias": 3.0, | |
| "lr": 100000000.0, | |
| "num_train_epochs": 1, | |
| "l1_strength": 0.0 | |
| } | |
| } |