Upload models/vit_beatrix_micro_rope_long/20250911_181250/best_epoch039_acc21.13_config.json with huggingface_hub
Browse files
models/vit_beatrix_micro_rope_long/20250911_181250/best_epoch039_acc21.13_config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_variant": "vit_beatrix_micro_rope_long",
|
| 3 |
+
"epoch": 39,
|
| 4 |
+
"val_acc": 21.130000006103515,
|
| 5 |
+
"session_timestamp": "20250911_181250",
|
| 6 |
+
"is_best": true,
|
| 7 |
+
"losses": {
|
| 8 |
+
"ce_loss": 3.4745617515563967,
|
| 9 |
+
"aux_loss": 0.0,
|
| 10 |
+
"geo_loss": 0.0,
|
| 11 |
+
"rose_loss": -0.434144322347641
|
| 12 |
+
},
|
| 13 |
+
"model_path": "best_epoch039_acc21.13.safetensors",
|
| 14 |
+
"optimizer_path": "best_epoch039_acc21.13_optimizer.pt",
|
| 15 |
+
"vocabulary_path": "best_epoch039_acc21.13_vocabulary.json",
|
| 16 |
+
"model_config": {
|
| 17 |
+
"variant": "vit_beatrix_micro_rope_long",
|
| 18 |
+
"vocab_dim": 256,
|
| 19 |
+
"embed_dim": 256,
|
| 20 |
+
"depth": 16,
|
| 21 |
+
"num_heads": 8,
|
| 22 |
+
"mlp_ratio": 4.0,
|
| 23 |
+
"img_size": 32,
|
| 24 |
+
"patch_size": 4,
|
| 25 |
+
"dropout": 0.0,
|
| 26 |
+
"attn_dropout": 0.0,
|
| 27 |
+
"similarity_mode": "rose",
|
| 28 |
+
"use_theta_head": true,
|
| 29 |
+
"theta_n_pentachora": 1024,
|
| 30 |
+
"theta_hidden_dim": 32,
|
| 31 |
+
"num_classes": 100
|
| 32 |
+
}
|
| 33 |
+
}
|