config.json

{
  "auto_map": {
    "AutoConfig": "configuration_finelap.FineLAPConfig",
    "AutoModel": "modeling_finelap.FineLAPModel"
  }, 
  "architectures": [
    "FineLAPModel"
  ],
  "audio_config": {
    "_attn_implementation_autoset": true,
    "activation_dropout": 0.0,
    "attn_drop_rate": 0.0,
    "depth": 12,
    "drop_rate": 0.0,
    "embed_dim": 768,
    "end_drop_path_rate": 0.0,
    "fixed_positions": true,
    "img_size": [
      1024,
      128
    ],
    "in_chans": 1,
    "layer_norm_first": false,
    "mel_bins": 128,
    "mlp_ratio": 4.0,
    "model_type": "eat",
    "model_variant": "pretrain",
    "norm_affine": true,
    "norm_eps": 1e-06,
    "num_classes": 527,
    "num_heads": 12,
    "patch_size": 16,
    "post_mlp_drop": 0.0,
    "qkv_bias": true,
    "start_drop_path_rate": 0.0,
    "stride": 16
  },
  "b_global": -10.0,
  "b_local": -10.0,
  "embed_size": 1024,
  "local_audio_proj_type": "transformer",
  "model_type": "finelap",
  "normalize_dense_audio_embeds": true,
  "temp_global": 0.1,
  "temp_local": 0.1,
  "text_encoder_name": "roberta-base",
  "torch_dtype": "float32",
  "transformers_version": "4.51.3",
  "unify_audio_proj": false
}