24 lines
736 B
JSON
24 lines
736 B
JSON
|
|
{
|
||
|
|
"run_name": "qwen4b_apparel23_bundle_sft_20260616_142947",
|
||
|
|
"work_root": "/var/tmp/qwen4b_apparel23_bundle_sft_20260616_142947",
|
||
|
|
"hf_dataset_repo": "flavianv/apparel23-qwen32b-kept-outfits-with-products",
|
||
|
|
"train_file": "/var/tmp/qwen4b_apparel23_bundle_sft_20260616_142947/data/apparel23_bundle_sft.train.jsonl",
|
||
|
|
"model_id": "Qwen/Qwen3-4B-Instruct-2507",
|
||
|
|
"out_dir": "/var/tmp/qwen4b_apparel23_bundle_sft_20260616_142947/model",
|
||
|
|
"learning_rate": 2e-05,
|
||
|
|
"epochs": 1.0,
|
||
|
|
"max_length": 768,
|
||
|
|
"per_device_train_batch_size": 1,
|
||
|
|
"gradient_accumulation_steps": 8,
|
||
|
|
"use_lora": false,
|
||
|
|
"lora_target_modules": [
|
||
|
|
"q_proj",
|
||
|
|
"k_proj",
|
||
|
|
"v_proj",
|
||
|
|
"o_proj",
|
||
|
|
"gate_proj",
|
||
|
|
"up_proj",
|
||
|
|
"down_proj"
|
||
|
|
]
|
||
|
|
}
|