| { | |
| "source_checkpoint_sha256": "d400c2e460e7287357f86083f84c187790a45b1d47985e908c52825373f888c3", | |
| "parent_release": "Grio43/OppaiOracle-experimental-X1", | |
| "parent_epoch": 21, | |
| "parent_step": 127974, | |
| "phase": 2, | |
| "epoch": 9, | |
| "step": 55719, | |
| "completed_at": "2026-09-17T01:43:41.793971", | |
| "model": { | |
| "architecture_type": "vit", | |
| "hidden_size": 896, | |
| "num_hidden_layers": 18, | |
| "num_attention_heads": 14, | |
| "intermediate_size": 2320, | |
| "image_size": 448, | |
| "patch_size": 16, | |
| "num_channels": 3, | |
| "hidden_dropout_prob": 0.0, | |
| "pos_dropout": 0.0, | |
| "attention_dropout": 0.0, | |
| "qk_norm": true, | |
| "layer_scale_init": 0.1, | |
| "drop_path_rate": 0.15, | |
| "initializer_range": 0.02, | |
| "layer_norm_eps": 1e-06, | |
| "use_fp32_layernorm": false, | |
| "use_flex_attention": true, | |
| "attention_bias": true, | |
| "flex_block_size": 128, | |
| "token_ignore_threshold": 0.9, | |
| "num_labels": 14996, | |
| "num_groups": 20, | |
| "tags_per_group": 10000, | |
| "gradient_checkpointing": false, | |
| "checkpoint_every_n_layers": 1 | |
| }, | |
| "training": { | |
| "memory_format": "contiguous", | |
| "num_epochs": 9, | |
| "learning_rate": 1.5e-05, | |
| "lr_scaling_mode": "sqrt", | |
| "lr_base_batch_size": 256, | |
| "weight_decay": 0.05, | |
| "gradient_accumulation_steps": 21, | |
| "world_size": 1, | |
| "optimizer": "adamw8bit", | |
| "adam_beta1": 0.9, | |
| "adam_beta2": 0.999, | |
| "adan_beta3": 0.99, | |
| "adam_epsilon": 1e-07, | |
| "scheduler": "wsd", | |
| "warmup_epochs": 1, | |
| "warmup_steps": null, | |
| "wsd_cooldown_fraction": 0.15, | |
| "wsd_plateau_enabled": false, | |
| "fp32_head_optimizer": true, | |
| "num_cycles": 1, | |
| "cycle_decay": 0.9, | |
| "lr_end": 0.0, | |
| "use_amp": true, | |
| "amp_dtype": "bfloat16", | |
| "enable_anomaly_detection": false, | |
| "max_grad_norm": 1.0, | |
| "gradient_clipping": { | |
| "enabled": true, | |
| "max_norm": 1.0 | |
| }, | |
| "save_steps": 1500, | |
| "save_total_limit": 3, | |
| "save_best_only": false, | |
| "resume_from": "latest", | |
| "eval_steps": 0, | |
| "logging_steps": 50, | |
| "resume_reseed": true, | |
| "allow_unverified_vocab_resume": false, | |
| "on_lr_schedule_change": "error", | |
| "tag_loss": { | |
| "alpha": 1.0, | |
| "gamma_neg": 3.0, | |
| "gamma_neg_override": null, | |
| "gamma_pos": 0.0, | |
| "detach_focal_weight": true, | |
| "label_smoothing": 0.0, | |
| "clip": 0.05, | |
| "class_weights": null, | |
| "class_weight_strategy": null, | |
| "class_weight_clip_min": 0.05, | |
| "class_weight_clip_max": 5.0, | |
| "class_weight_beta": 0.9999 | |
| }, | |
| "phase": 2, | |
| "asl_schedule": { | |
| "enabled": false, | |
| "min_dwell_epochs": 3, | |
| "phase1": { | |
| "gamma_neg_min": 3.0, | |
| "gamma_neg_max": 3.0, | |
| "hold_epochs": 8 | |
| }, | |
| "phase2": { | |
| "gamma_neg_min": 3.0, | |
| "gamma_neg_max": 3.0, | |
| "hold_epochs": 2 | |
| }, | |
| "phase3": { | |
| "gamma_neg_min": 3.0, | |
| "gamma_neg_max": 3.0, | |
| "hold_epochs": 0 | |
| } | |
| }, | |
| "asl_telemetry": { | |
| "enabled": true, | |
| "interval_updates": 100, | |
| "log_every_updates": 500, | |
| "ema_beta": 0.98, | |
| "topk_hard": 10, | |
| "num_deciles": 10, | |
| "hist_min": 0.05, | |
| "hist_max": 0.95, | |
| "hist_bins": 18, | |
| "watch_band_low": 0.2, | |
| "watch_band_high": 0.5, | |
| "epr_alarm_rel_drop": 0.05, | |
| "epr_alarm_window_epochs": 2, | |
| "exclude_rating_tags": true, | |
| "sibling_groups_path": "./configs/confusable_groups.json", | |
| "shadow_controller_enabled": true, | |
| "shadow_lambda": 0.05, | |
| "shadow_delta_p_target": 0.2 | |
| }, | |
| "device": "cuda", | |
| "use_compile": true, | |
| "compile_mode": "max-autotune-no-cudagraphs", | |
| "compile_fullgraph": false, | |
| "compile_dynamic": true, | |
| "use_tensorboard": true, | |
| "seed": 42, | |
| "deterministic": false, | |
| "benchmark": true, | |
| "selection_metric": "val_mAP", | |
| "early_stopping_patience": 4, | |
| "early_stopping_threshold": 5e-07, | |
| "early_stopping_burn_in_epochs": 2, | |
| "early_stopping_burn_in_strategy": "median", | |
| "early_stopping_mode": "advise", | |
| "early_stopping_min_delta": 0.003, | |
| "early_stopping_plateau_slope_threshold": 0.0001, | |
| "early_stopping_plateau_noise_multiplier": 2.0, | |
| "early_stopping_window": 5, | |
| "early_stopping_confirm": 2, | |
| "early_stopping_min_epochs": 10, | |
| "early_stopping_regression_delta": 0.006, | |
| "early_stopping_regression_confirm": 2, | |
| "early_stopping_overfit_confirm": 3, | |
| "early_stopping_overfit_window": 8, | |
| "early_stopping_overfit_min_relative_change": 0.01, | |
| "early_stopping_overfit_noise_multiplier": 2.0, | |
| "early_stopping_overfit_ranking_tolerance": 0.0001, | |
| "wsd_deterioration_actions": { | |
| "peak_decline": "preserve", | |
| "overfit_loss": "warn" | |
| }, | |
| "early_stopping_phases": { | |
| "phase1": { | |
| "mode": null, | |
| "window": 5, | |
| "confirm": 2, | |
| "min_epochs": 10, | |
| "plateau_slope_threshold": null, | |
| "plateau_noise_multiplier": null | |
| }, | |
| "phase2": { | |
| "mode": "advise", | |
| "window": 5, | |
| "confirm": 2, | |
| "min_epochs": 10, | |
| "plateau_slope_threshold": null, | |
| "plateau_noise_multiplier": null | |
| }, | |
| "phase3": { | |
| "mode": "off", | |
| "window": null, | |
| "confirm": null, | |
| "min_epochs": null, | |
| "plateau_slope_threshold": null, | |
| "plateau_noise_multiplier": null | |
| } | |
| }, | |
| "use_distillation": false, | |
| "distillation_alpha": 0.7, | |
| "distillation_temperature": 3.0 | |
| }, | |
| "data_recipe": { | |
| "vocab_min_frequency": 550, | |
| "vocab_min_frequency_by_category": { | |
| "artist": 550, | |
| "char": 550, | |
| "copyright": 550, | |
| "gen": 450, | |
| "meta": 550 | |
| }, | |
| "random_flip_prob": 0.5, | |
| "horizontal_flip_exclude_tags": [ | |
| "gen:left-handed", | |
| "gen:left-to-right_manga", | |
| "gen:right-over-left_kimono", | |
| "gen:right-to-left_comic" | |
| ], | |
| "color_jitter_enabled": true, | |
| "color_jitter_brightness": 0.22, | |
| "color_jitter_brightness_p": 0.5, | |
| "color_jitter_contrast": 0.15, | |
| "color_jitter_contrast_p": 0.5, | |
| "color_jitter_saturation": 0.06, | |
| "color_jitter_saturation_p": 0.5, | |
| "random_erasing_enabled": false, | |
| "random_erasing_p": 0.35, | |
| "random_erasing_scale_min": 0.007, | |
| "random_erasing_scale_max": 0.03, | |
| "random_erasing_ratio_min": 0.3, | |
| "random_erasing_ratio_max": 2.0, | |
| "random_rotation_enabled": true, | |
| "random_rotation_p": 0.3, | |
| "random_rotation_min_degrees": 2.0, | |
| "random_rotation_max_degrees": 5.0, | |
| "gaussian_blur_enabled": true, | |
| "gaussian_blur_p": 0.15, | |
| "gaussian_blur_kernel_size": 3, | |
| "gaussian_blur_sigma_min": 0.1, | |
| "gaussian_blur_sigma_max": 1.0 | |
| }, | |
| "train_images": 6239823, | |
| "eval_images": 30000, | |
| "dataset_maintenance": { | |
| "danbooru_metadata_sync": "2026-08", | |
| "deprecated_tags_removed": true, | |
| "renamed_tags_updated_to_current_names": true, | |
| "provenance": "model author" | |
| }, | |
| "metrics": { | |
| "train_loss": 0.0014653843827545643, | |
| "val_loss": 0.0016158618964254856, | |
| "val_f1_macro": 0.45742398500442505, | |
| "val_f1_micro": 0.6621212959289551, | |
| "val_mAP": 0.5750626921653748, | |
| "selection_metric": "val_mAP" | |
| }, | |
| "scheduler": { | |
| "warmup_steps": 6191, | |
| "total_steps": 55719, | |
| "cooldown_fraction": 0.15, | |
| "budget_cooldown_start": 48451, | |
| "min_lr": 0.0, | |
| "base_max_lr": 2.9764702249476646e-05, | |
| "cooldown_start": 48451, | |
| "cooldown_steps": 7268, | |
| "cooldown_reason": "budget", | |
| "cooldown_best_metric": 0.5750626921653748, | |
| "base_lrs": [ | |
| 2.9764702249476646e-05, | |
| 2.9764702249476646e-05 | |
| ], | |
| "last_epoch": 55719, | |
| "_step_count": 55720, | |
| "_is_initial": false, | |
| "_get_lr_called_within_step": false, | |
| "_last_lr": [ | |
| 0.0, | |
| 0.0 | |
| ] | |
| } | |
| } | |