best_model.json

{
  "model_name": "David-partial_shared-deep_efficiency",
  "run_id": "20251012_161107",
  "timestamp": "2025-10-12T16:35:56.107463",
  "best_val_acc": 84.586,
  "best_epoch": 6,
  "final_train_acc": 93.03353895315755,
  "final_train_loss": 0.6332142021233281,
  "scale_accuracies": {
    "384": 84.672,
    "512": 84.618,
    "768": 84.586,
    "1024": 84.368,
    "1280": 84.366,
    "1536": 84.46,
    "1792": 84.254,
    "2048": 84.364
  },
  "architecture": {
    "preset": "clip_vit_bigg14",
    "sharing_mode": "partial_shared",
    "fusion_mode": "deep_efficiency",
    "scales": [
      384,
      512,
      768,
      1024,
      1280,
      1536,
      1792,
      2048
    ],
    "feature_dim": 1280,
    "num_classes": 1000,
    "use_belly": true,
    "belly_expand": 2.0
  },
  "training": {
    "dataset": "AbstractPhil/imagenet-clip-features-orderly",
    "model_variant": "clip_vit_laion_bigg14",
    "num_epochs": 10,
    "batch_size": 1024,
    "learning_rate": 0.001,
    "rose_weight": "0.1\u21920.5",
    "cayley_loss": false,
    "optimizer": "AdamW",
    "scheduler": "cosine_restarts"
  },
  "files": {
    "weights_safetensors": "weights/David-partial_shared-deep_efficiency/20251012_161107/best_model_acc84.59.safetensors",
    "weights_pytorch": "weights/David-partial_shared-deep_efficiency/20251012_161107/best_model.pth",
    "config": "weights/David-partial_shared-deep_efficiency/20251012_161107/david_config.json",
    "training_config": "weights/David-partial_shared-deep_efficiency/20251012_161107/train_config.json",
    "tensorboard": "runs/David-partial_shared-deep_efficiency/20251012_161107/"
  }
}