tokenizer/config.json

{
  "_name_or_path": "openai/whisper-large-v3",
  "activation_dropout": 0.0,
  "activation_function": "gelu",
  "apply_spec_augment": false,
  "architectures": [
    "WhisperLFQEncoder"
  ],
  "attention_dropout": 0.0,
  "batch_maximization_weight": 0.3,
  "classifier_proj_size": 256,
  "d_model": 1280,
  "delta_cross_logdet_reg_loss": 1.0,
  "dropout": 0.0,
  "enable_noise_contrastive": false,
  "encoder_attention_heads": 20,
  "encoder_causal_attention": false,
  "encoder_causal_convolution": false,
  "encoder_ffn_dim": 5120,
  "encoder_layerdrop": 0.0,
  "encoder_layers": 32,
  "fix_noise_commit_loss_ratio": false,
  "init_std": 0.06,
  "is_encoder_decoder": true,
  "lambda_consensus_loss_ratio": 0.25,
  "layernorm_after_quantize": true,
  "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,
  "max_length": 448,
  "max_source_positions": 1500,
  "max_target_positions": 448,
  "median_filter_width": 7,
  "model_type": "whisper",
  "noise_commit_loss_ratio": 0.3333333333333333,
  "num_clean_input": 3,
  "num_hidden_layers": 32,
  "num_mel_bins": 128,
  "num_voters": 5,
  "pooling_kernel_size": 2,
  "pooling_position": 16,
  "pooling_type": "avg",
  "quantize_commit_coefficient": 0.25,
  "quantize_encoder_only": true,
  "quantize_position": 16,
  "quantize_vocab_size": 8192,
  "sample_minimization_weight": 0.1,
  "torch_dtype": "float32",
  "transformers_version": "4.48.0",
  "use_aggregated_entropy": true,
  "use_cache": true,
  "use_codebook_ce_loss": true,
  "use_commit_loss": true,
  "use_projection_bias": false,
  "use_svlfq": true,
  "use_weighted_layer_sum": false
}