{
  "model_type": "averformer",
  "architecture": "AVERFormer",
  "num_classes": 8,
  "emotions": [
    "neutral",
    "calm",
    "happy",
    "sad",
    "angry",
    "fearful",
    "disgust",
    "surprised"
  ],
  "visual_backbone": "uniformer_small",
  "audio": {
    "sample_rate": 22000,
    "duration_sec": 4.0,
    "channels": 2,
    "n_fft": 400,
    "hop_length": 257,
    "n_mels_per_channel": 160,
    "n_mfcc_per_channel": 20,
    "target_time_frames": 343
  },
  "video": {
    "num_frames": 12,
    "height": 180,
    "width": 320,
    "imagenet_mean": [
      0.485,
      0.456,
      0.406
    ],
    "imagenet_std": [
      0.229,
      0.224,
      0.225
    ]
  },
  "source_checkpoint": "best_model_epoch_161.pth"
}