Upload models/vit_zana_nano_thicc/20250916_221953/best_epoch040_acc52.21_training_config.json with huggingface_hub
Browse files
models/vit_zana_nano_thicc/20250916_221953/best_epoch040_acc52.21_training_config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"apply_margin_train_only": false,
|
| 3 |
+
"attn_dropout": 0.0,
|
| 4 |
+
"batch_size": 256,
|
| 5 |
+
"checkpoint_path": null,
|
| 6 |
+
"continue_last": false,
|
| 7 |
+
"continue_optimizer": true,
|
| 8 |
+
"continue_scheduler": true,
|
| 9 |
+
"cutmix_alpha": 1.0,
|
| 10 |
+
"depth": 4,
|
| 11 |
+
"dropout": 0.0,
|
| 12 |
+
"embed_dim": 128,
|
| 13 |
+
"epochs": 200,
|
| 14 |
+
"head_type": "roseface",
|
| 15 |
+
"hub_repo": "AbstractPhil/penta-vit-experiments",
|
| 16 |
+
"hub_token": null,
|
| 17 |
+
"img_size": 32,
|
| 18 |
+
"label_smoothing": 0.1,
|
| 19 |
+
"log_dir": "./runs",
|
| 20 |
+
"lr": 0.001,
|
| 21 |
+
"margin_m": 0.3,
|
| 22 |
+
"margin_type": "cosface",
|
| 23 |
+
"mix_final_alpha": 0.5,
|
| 24 |
+
"mix_final_prob": 0.25,
|
| 25 |
+
"mix_prob": 0.01,
|
| 26 |
+
"mix_schedule": "cosine",
|
| 27 |
+
"mix_schedule_end": null,
|
| 28 |
+
"mix_schedule_start": 0,
|
| 29 |
+
"mixup_alpha": 0.8,
|
| 30 |
+
"mlp_ratio": 4.0,
|
| 31 |
+
"model_variant": "vit_zana_nano_thicc",
|
| 32 |
+
"norm_type": "l1",
|
| 33 |
+
"num_heads": 8,
|
| 34 |
+
"patch_size": 2,
|
| 35 |
+
"prototype_mode": "centroid",
|
| 36 |
+
"push_every": 20,
|
| 37 |
+
"push_to_hub": true,
|
| 38 |
+
"save_dir": "./checkpoints/pentachora",
|
| 39 |
+
"save_every": 10,
|
| 40 |
+
"scale_s": 30.0,
|
| 41 |
+
"session_timestamp": "20250916_221953",
|
| 42 |
+
"similarity_mode": "rose",
|
| 43 |
+
"vocab_dim": 128,
|
| 44 |
+
"warmup_epochs": 10,
|
| 45 |
+
"weight_decay": 1e-05
|
| 46 |
+
}
|