| { |
| "timestamp": "2026-03-08T08:45:18.769911", |
| "command_line_args": { |
| "base_model": "Qwen/Qwen3-4B-Base", |
| "ft_model": "/work8/yizhanh/0444_global_step_285_hf", |
| "save_path": "/work8/yizhanh/mhm/MergeBench/math_rl", |
| "task_name": "math", |
| "sparsity": 0.1, |
| "sigmoid_bias": 3.0, |
| "lr": 100000000.0, |
| "n_epochs": 1, |
| "l1_strength": 0.0, |
| "gpu_id": 5, |
| "sample_size": 100 |
| }, |
| "actual_results": { |
| "actual_sparsity": 0.1026563279741487, |
| "target_sparsity": 0.1, |
| "mask_proportion": 0.1026563279741487, |
| "dataset_size": 100 |
| }, |
| "task_similarity": { |
| "masked_energy_ratio": 0.9903082846604907, |
| "cosine_similarity_in_mask_region": -2.3204116587294266e-05, |
| "mask_coverage_ratio": 0.1026563279741487, |
| "task_vector_l2_norm": 1.0, |
| "masked_l2_norm": 0.9903082847595215, |
| "unmasked_l2_norm": 0.43302464485168457 |
| }, |
| "graft_args": { |
| "sparsity": 0.1, |
| "sigmoid_bias": 3.0, |
| "lr": 100000000.0, |
| "num_train_epochs": 1, |
| "l1_strength": 0.0 |
| }, |
| "task_similarity_between_models": { |
| "jaccard_similarity": 0.369397455060345, |
| "intersection_size": 306181056, |
| "union_size": 828866176, |
| "mask_1_size": 373003008, |
| "mask_2_size": 762044288, |
| "intersection_ratio_mask1": 0.8208541202970674, |
| "intersection_ratio_mask2": 0.4017890571735379, |
| "avg_intersection_ratio": 0.6113215887353026 |
| } |
| } |
