aves_especies/aves-base-bio.torchaudio.model_config.json

{
    "extractor_mode": "group_norm",
    "extractor_conv_layer_config": [
        [
            512,
            10,
            5
        ],
        [
            512,
            3,
            2
        ],
        [
            512,
            3,
            2
        ],
        [
            512,
            3,
            2
        ],
        [
            512,
            3,
            2
        ],
        [
            512,
            2,
            2
        ],
        [
            512,
            2,
            2
        ]
    ],
    "extractor_conv_bias": false,
    "encoder_embed_dim": 768,
    "encoder_projection_dropout": 0.1,
    "encoder_pos_conv_kernel": 128,
    "encoder_pos_conv_groups": 16,
    "encoder_num_layers": 12,
    "encoder_num_heads": 12,
    "encoder_attention_dropout": 0.1,
    "encoder_ff_interm_features": 3072,
    "encoder_ff_interm_dropout": 0.0,
    "encoder_dropout": 0.1,
    "encoder_layer_norm_first": false,
    "encoder_layer_drop": 0.05
}