{
  "architectures": [
    "CohereAsrForConditionalGeneration"
  ],
  "attention_bias": true,
  "attention_dropout": 0.0,
  "bos_token_id": 4,
  "decoder_start_token_id": 85,
  "dtype": "float32",
  "encoder_config": {
    "activation_dropout": 0.0,
    "attention_bias": true,
    "attention_dropout": 0.0,
    "conv_kernel_size": 9,
    "convolution_bias": true,
    "dropout": 0.0,
    "dropout_positions": 0.0,
    "hidden_act": "silu",
    "hidden_size": 16,
    "initializer_range": 0.02,
    "intermediate_size": 32,
    "layerdrop": 0.0,
    "max_position_embeddings": 5000,
    "model_type": "parakeet_encoder",
    "num_attention_heads": 2,
    "num_hidden_layers": 2,
    "num_key_value_heads": 2,
    "num_mel_bins": 8,
    "scale_input": false,
    "subsampling_conv_channels": 8,
    "subsampling_conv_kernel_size": 3,
    "subsampling_conv_stride": 2,
    "subsampling_factor": 4
  },
  "eos_token_id": 3,
  "head_dim": 8,
  "hidden_act": "relu",
  "hidden_size": 16,
  "initializer_range": 0.02,
  "intermediate_size": 32,
  "is_encoder_decoder": true,
  "max_position_embeddings": 1024,
  "model_type": "cohere_asr",
  "num_attention_heads": 2,
  "num_hidden_layers": 2,
  "num_key_value_heads": 2,
  "pad_token_id": 2,
  "tie_word_embeddings": false,
  "transformers_version": "5.16.0.dev0",
  "use_cache": true,
  "vocab_size": 16384
}