{
  "model_type": "mhubert_ipa_ctc_ft",
  "architectures": [
    "MHuBERTIPACTCFTModel"
  ],
  "auto_map": {
    "AutoConfig": "configuration_mhubert_ipa_ctc_ft.MHuBERTIPACTCFTConfig",
    "AutoModel": "modeling_mhubert_ipa_ctc_ft.MHuBERTIPACTCFTModel"
  },
  "base_model": "utter-project/mHuBERT-147",
  "backbone_id": "utter-project/mHuBERT-147",
  "task": "end-to-end CTC IPA phone recognition",
  "label_type": "IPA",
  "label_map": "ipa_map.json",
  "ctc_blank_token": "<blank>",
  "blank_position": "last",
  "architecture": {
    "input_dim": 768,
    "proj_dim": 256,
    "lstm_hidden": 256,
    "lstm_layers": 2,
    "lstm_bidirectional": true,
    "dropout": 0.3,
    "n_phones": 45,
    "output_dim": 46,
    "blank_id": 45
  },
  "evaluation": {
    "timit_test_per": 0.0896,
    "buckeye_val_per": 0.1987,
    "metric": "PER (Phone Error Rate)"
  },
  "backbone_config": {
    "activation_dropout": 0.1,
    "apply_spec_augment": true,
    "attention_dropout": 0.1,
    "classifier_proj_size": 256,
    "conv_bias": false,
    "conv_dim": [
      512,
      512,
      512,
      512,
      512,
      512,
      512
    ],
    "conv_kernel": [
      10,
      3,
      3,
      3,
      3,
      2,
      2
    ],
    "conv_stride": [
      5,
      2,
      2,
      2,
      2,
      2,
      2
    ],
    "do_stable_layer_norm": false,
    "feat_extract_activation": "gelu",
    "feat_extract_dropout": 0.0,
    "feat_extract_norm": "group",
    "feat_proj_dropout": 0.1,
    "feat_proj_layer_norm": true,
    "final_dropout": 0.1,
    "gradient_checkpointing": false,
    "hidden_act": "gelu",
    "hidden_dropout": 0.1,
    "hidden_dropout_prob": 0.1,
    "hidden_size": 768,
    "initializer_range": 0.02,
    "intermediate_size": 3072,
    "layer_norm_eps": 1e-05,
    "layerdrop": 0.1,
    "mask_feature_length": 10,
    "mask_feature_min_masks": 0,
    "mask_feature_prob": 0.0,
    "mask_time_length": 10,
    "mask_time_min_masks": 2,
    "mask_time_prob": 0.05,
    "model_type": "hubert",
    "num_attention_heads": 12,
    "num_conv_pos_embedding_groups": 16,
    "num_conv_pos_embeddings": 128,
    "num_feat_extract_layers": 7,
    "num_hidden_layers": 12,
    "torch_dtype": "float32",
    "use_weighted_layer_sum": false,
    "vocab_size": 32
  }
}