28 lines
1.2 KiB
JSON
28 lines
1.2 KiB
JSON
|
{
|
|||
|
|
"format": "nano-v3.1-multi",
|
|||
|
|
"base_model_id": "Qwen/Qwen2.5-14B-Instruct",
|
|||
|
|
"hidden_layers": 48,
|
|||
|
|
"locked_count": 76,
|
|||
|
|
"pending_8bit": 260,
|
|||
|
|
"elapsed_seconds": 5959,
|
|||
|
|
"build_reference_mode": "8bit",
|
|||
|
|
"reference_scope": "original_baseline",
|
|||
|
|
"pending_policy": "leave_in_base_8bit",
|
|||
|
|
"pruned_from": "/kaggle/working/final_artifact_Qwen2.5-14B-Instruct",
|
|||
|
|
"pruned_strategy": "remove_last_locked",
|
|||
|
|
"pruned_removed_count": 9,
|
|||
|
|
"pruned_removed_modules": [
|
|||
|
|
"model.layers.42.mlp.down_proj",
|
|||
|
|
"model.layers.45.mlp.gate_proj",
|
|||
|
|
"model.layers.46.mlp.down_proj",
|
|||
|
|
"model.layers.28.self_attn.v_proj",
|
|||
|
|
"model.layers.29.mlp.down_proj",
|
|||
|
|
"model.layers.34.mlp.gate_proj",
|
|||
|
|
"model.layers.36.mlp.up_proj",
|
|||
|
|
"model.layers.42.self_attn.k_proj",
|
|||
|
|
"model.layers.46.self_attn.o_proj"
|
|||
|
|
],
|
|||
|
|
"self_contained": true,
|
|||
|
|
"base_model_local_subdir": "."
|
|||
|
|
}
|