| batch_size_training: '32' | |
| checkpoint_type: StateDictType.FULL_STATE_DICT | |
| dataset: rosft_dataset | |
| dist_checkpoint_folder: fine-tuned | |
| dist_checkpoint_root_folder: test_run_save-035-V3 | |
| enable_fsdp: 'True' | |
| freeze_layers: 'False' | |
| fsdp_activation_checkpointing: 'True' | |
| gamma: '0.9' | |
| load_peft_model: 'False' | |
| low_cpu_fsdp: 'False' | |
| lr: 5e-05 | |
| micro_batch_size: '32' | |
| mixed_precision: 'True' | |
| model_name: models/v3/llama7b-full-1e-4_low-chunk1024-018-035 | |
| num_epochs: '1' | |
| num_freeze_layers: '1' | |
| num_workers_dataloader: '2' | |
| one_gpu: 'False' | |
| optimizer: AdamW | |
| output_dir: PATH/to/save/PEFT/model | |
| peft_method: lora | |
| pure_bf16: 'True' | |
| quantization: 'False' | |
| run_validation: 'True' | |
| save_model: 'True' | |
| save_optimizer: 'False' | |
| seed: '42' | |
| sharding_strategy: ShardingStrategy.FULL_SHARD | |
| type_of_model: chat | |
| use_fast_kernels: 'False' | |
| use_fp16: 'False' | |
| use_peft: 'False' | |
| val_batch_size: '64' | |
| weight_decay: '0.0' | |