config.json

{
  "model_type": "averformer-v4",
  "architecture": "AVERFormerV4",
  "corpus": "MELD",
  "num_classes": 7,
  "emotions": [
    "neutral",
    "joy",
    "sadness",
    "anger",
    "fear",
    "disgust",
    "surprise"
  ],
  "audio_backbone": "microsoft/wavlm-large",
  "video_backbone": "MCG-NJU/videomae-large",
  "text_backbone": "microsoft/deberta-v3-large",
  "fusion_layers": 2,
  "lora_r": 16,
  "unfreeze_audio": 0,
  "unfreeze_video": 0,
  "unfreeze_text": 0,
  "fusion_dim": 512,
  "use_text": true,
  "audio_format": {
    "sample_rate": 16000,
    "channels": 1,
    "expected_seconds": "4.0 (rolling buffer; flexible)"
  },
  "video_format": {
    "num_frames": 16,
    "height": 224,
    "width": 224,
    "imagenet_mean": [
      0.485,
      0.456,
      0.406
    ],
    "imagenet_std": [
      0.229,
      0.224,
      0.225
    ],
    "value_range": "[0, 1]"
  },
  "text_format": {
    "tokenizer": "microsoft/deberta-v3-large",
    "max_length": 256,
    "n_ctx_turns": 3,
    "speaker_embedding": true
  },
  "select_metric": "wF1",
  "source_checkpoint": "pytorch_model.pth",
  "training_script": "train_v5_spec.py / train_v6_supcon.py / train_v10_mcncl.py",
  "val_score": 0.3815709082075283,
  "best_epoch": 4,
  "seed": 1337
}