| { |
| "dataset": "cifar10", |
| "img": 32, |
| "patch": 4, |
| "in_ch": 3, |
| "dim": 384, |
| "depth": 12, |
| "heads": 6, |
| "num_classes": 10, |
| "mlp_ratio": 4, |
| "T": 1000, |
| "batch": 128, |
| "lr": 0.0001, |
| "ema_decay": 0.9999, |
| "lam_energy": 0.002, |
| "tau": 1.0, |
| "class_dropout": 0.1, |
| "target_keep": 0.5, |
| "always_on": 2, |
| "grad_clip": 1.0, |
| "run_tag": "baseline", |
| "total_steps": 200000, |
| "warmup_energy_steps": 3000, |
| "base_seed": 1234, |
| "ckpt_every_steps": 1000, |
| "push_interval_min": 20, |
| "min_push_gap_min": 5, |
| "disk_cap_gb": 17.0, |
| "hf_user": "Shanmuk4622", |
| "device": "cuda", |
| "amp": true, |
| "grad_ckpt": false, |
| "num_workers": 2, |
| "smoke": false, |
| "use_head_axis": false, |
| "calibrated_cost_repo": null |
| } |