Views
No views yet
N/Asparsed_wudi_merge/workspace/outs/wudi_sparsed_0.81{
2 "out": "/workspace/outs/wudi_sparsed_0.8",
3 "scaling": 1.0,
4 "dtype": "bfloat16",
5 "device_map": "cpu",
6 "merge_method": "sparsed_wudi_merge",
7 "exclude": [
8 ".*embed.*",
9 ".*lm_head.*"
10 ],
11 "wudi_variant": "wudi_all_linear",
12 "wudi_iter": 200,
13 "wudi_lr": 1e-05,
14 "wudi_weight_decay": 0.0,
15 "wudi_device": "cuda",
16 "wudi_fallback": "task_arithmetic",
17 "wudi_fallback_scaling": 1.0,
18 "wudi_K": 0.8,
19 "effective_exclude": [
20 ".*embed_tokens\\.weight$",
21 ".*lm_head\\.weight$",
22 ".*embed.*",
23 ".*lm_head.*"
24 ]
25}