DeepXR/Helion-V1
417
1{2 "output_dir": "./helion-v1-checkpoints",3 "num_train_epochs": 3,4 "per_device_train_batch_size": 4,5 "per_device_eval_batch_size": 4,6 "gradient_accumulation_steps": 8,7 "eval_strategy": "steps",8 "eval_steps": 500,9 "save_strategy": "steps",10 "save_steps": 500,11 "save_total_limit": 3,12 "learning_rate": 2e-5,13 "warmup_steps": 100,14 "logging_steps": 10,15 "logging_dir": "./logs",16 "fp16": false,17 "bf16": true,18 "optim": "adamw_torch",19 "weight_decay": 0.01,20 "lr_scheduler_type": "cosine",21 "max_grad_norm": 1.0,22 "gradient_checkpointing": true,23 "gradient_checkpointing_kwargs": {24 "use_reentrant": false25 },26 "dataloader_num_workers": 4,27 "group_by_length": true,28 "report_to": ["tensorboard", "wandb"],29 "load_best_model_at_end": true,30 "metric_for_best_model": "eval_loss",31 "greater_is_better": false,32 "ddp_find_unused_parameters": false,33 "max_seq_length": 4096,34 "packing": false35}