File size: 1,317 Bytes
36c201b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 | {
"base_model": "<LOCAL_PATH>/Qwen3-8B",
"task": "metamath",
"peft_method": "lora",
"output_dir": "<LOCAL_PATH>/metamath",
"dataset_path": "<LOCAL_PATH>/metamathqa-train.parquet",
"train_profile": "none",
"force_max_steps": false,
"max_steps": null,
"num_train_epochs": 3.0,
"stop_at_step": null,
"stop_at_epoch": null,
"resume_from_checkpoint": null,
"per_device_train_batch_size": 16,
"gradient_accumulation_steps": 2,
"global_train_batch_size": 32,
"lr": 0.0001,
"warmup_ratio": 0.05,
"weight_decay": 0.0,
"min_lr_ratio": null,
"lr_scheduler_type": "cosine",
"adam_beta1": 0.9,
"adam_beta2": 0.999,
"grad_clip": 1.0,
"max_seq_len": 4096,
"seed": 44,
"save_strategy": "epoch",
"save_steps": 500,
"save_total_limit": 1,
"logging_steps": 25,
"max_train_samples": 50000,
"dataset_seed": 42,
"bf16": true,
"fp16": false,
"gradient_checkpointing": true,
"group_by_length": false,
"pad_to_multiple_of": 8,
"dataloader_num_workers": 0,
"optim": null,
"use_qlora": false,
"sft_format": "chat",
"chat_template_mode": "non_thinking",
"target_modules": "all",
"r": 16,
"lora_alpha": 32,
"lora_dropout": 0.05,
"pissa_init_mode": "pissa",
"init_r": 32,
"target_r": 16,
"loraplus_lr_ratio": 20.0,
"loraplus_lr_embedding": null
}
|