16 lines
628 B
JSON
16 lines
628 B
JSON
|
{
|
||
|
|
"model_id": "Qwen/Qwen3-0.6B-Base",
|
||
|
|
"run_name": "qwen3-0.6b-math-l45-qlora-sft-v2",
|
||
|
|
"hf_repo_id": "zannvznn/qwen3-0.6b-math-l45-qlora-merged-fp16-v2",
|
||
|
|
"adapter_dir": "/kaggle/working/qwen3-0.6b-math-l45-qlora-sft-v2/adapter",
|
||
|
|
"merged_dir": "/kaggle/working/qwen3-0.6b-math-l45-qlora-sft-v2/merged-fp16",
|
||
|
|
"merge_method": "manual_lora_cpu",
|
||
|
|
"merged_matrix_count": 196,
|
||
|
|
"unique_target_module_count": 196,
|
||
|
|
"elapsed_seconds": 0.4837789409994002,
|
||
|
|
"source_adapter_dir": "/kaggle/working/qwen3-0.6b-math-l45-qlora-sft-v2/adapter",
|
||
|
|
"lora_alpha": 32.0,
|
||
|
|
"r": 16,
|
||
|
|
"use_rslora": false,
|
||
|
|
"fan_in_fan_out": false
|
||
|
|
}
|