deepcoder2024's picture
Add LoRA adapter weights and model card for Cochrane screening
25ca8d6 verified
Raw
History Blame Contribute Delete
648 Bytes
{
"model_path": "Qwen/Qwen3-4B",
"train_file": "deepcoder2024/cochrane-screening-sft (train)",
"val_file": "deepcoder2024/cochrane-screening-sft (validation)",
"output_dir": "Qwen3-4B-LoRA-Cochrane-Screening",
"max_length": 2048,
"num_train_epochs": 1.0,
"learning_rate": 0.0002,
"per_device_train_batch_size": 1,
"per_device_eval_batch_size": 1,
"gradient_accumulation_steps": 16,
"warmup_ratio": 0.03,
"weight_decay": 0.01,
"logging_steps": 50,
"eval_steps": 2000,
"save_steps": 2000,
"save_total_limit": 2,
"lora_r": 16,
"lora_alpha": 32,
"lora_dropout": 0.05,
"seed": 42,
"dataloader_num_workers": 2
}