| { | |
| "dataset_name": "data/packaged_pretrain_dataset.parquet", | |
| "num_proc": 1, | |
| "max_seq_length": 32, | |
| "seed": 0, | |
| "optim": "adamw_torch", | |
| "max_steps": 60, | |
| "per_device_train_batch_size": 2, | |
| "learning_rate": 1e-05, | |
| "weight_decay": 0, | |
| "warmup_steps": 10, | |
| "lr_scheduler_type": "linear", | |
| "gradient_checkpointing": true, | |
| "dataloader_num_workers": 2, | |
| "bf16": true, | |
| "gradient_accumulation_steps": 1, | |
| "logging_steps": 3, | |
| "report_to": [], | |
| "save_strategy": "steps", | |
| "save_steps": 3, | |
| "save_total_limit": 1, | |
| "push_to_hub": true, | |
| "hub_model_id": "hienbm/psychology-llama3.1-8B" | |
| } |