| { | |
| "learning_rate": 0.0001, | |
| "weight_decay": 0.01, | |
| "betas": [ | |
| 0.9, | |
| 0.95 | |
| ], | |
| "grad_clip": 1.0, | |
| "optimizer": "adamw", | |
| "warmup_steps": 500, | |
| "total_steps": 417059, | |
| "max_epochs": 100, | |
| "lr_scheduler": "constant", | |
| "min_lr": 1e-06, | |
| "warmup_type": "linear", | |
| "gradient_accumulation": 8, | |
| "gradient_checkpointing": false, | |
| "compile_mode": null, | |
| "ema_decay": 0.9999, | |
| "shift": 3.0, | |
| "use_snr_weighting": true, | |
| "snr_gamma": 5.0, | |
| "use_huber_loss": true, | |
| "huber_delta": 0.1, | |
| "logit_normal_sampling": true, | |
| "logit_mean": 0.0, | |
| "logit_std": 1.0, | |
| "enable_lune": true, | |
| "lune_weight": 0.1, | |
| "lune_warmup_steps": 1000, | |
| "lune_dropout": 0.1, | |
| "lune_mode": "cosine", | |
| "enable_sol": true, | |
| "sol_weight": 0.05, | |
| "sol_warmup_steps": 2000, | |
| "sol_dropout": 0.1, | |
| "use_spatial_weighting": false, | |
| "text_dropout": 0.1, | |
| "checkpoint_dir": "./checkpoints", | |
| "save_every_steps": 1562, | |
| "keep_last_n_steps": 5, | |
| "save_every_epochs": 1, | |
| "keep_last_n_epochs": 3, | |
| "log_every": 20, | |
| "tensorboard_dir": "./logs", | |
| "sample_every": 500, | |
| "sample_prompts": [ | |
| "subject, object, a red cube on a blue sphere", | |
| "subject, a golden retriever sitting on grass", | |
| "subject, object, a coffee cup next to a laptop", | |
| "subject, a woman with short hair smiling" | |
| ], | |
| "sample_dir": "./samples", | |
| "dtype": "torch.bfloat16", | |
| "hf_repo_id": "AbstractPhil/tiny-flux-deep", | |
| "upload_every_steps": 1562, | |
| "upload_every_epochs": 0 | |
| } |