Add EmformerProcessor, clean up the config

config.json

@@ -1,92 +1,35 @@
 {
   "activation_dropout": 0.1,
-  "apply_spec_augment": true,
   "architectures": [
     "EmformerForRNNT"
   ],
   "attention_dropout": 0.1,
-  "bos_token_id": 1,
-  "codevector_dim": 256,
-  "contrastive_logits_temperature": 0.1,
-  "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
-  ],
-  "ctc_loss_reduction": "sum",
-  "ctc_zero_infinity": false,
-  "diversity_loss_weight": 0.1,
-  "do_stable_layer_norm": false,
-  "eos_token_id": 2,
-  "feat_extract_activation": "gelu",
-  "feat_extract_norm": "group",
-  "feat_proj_dropout": 0.0,
-  "feat_quantizer_dropout": 0.0,
-  "ffn_dim": 2048,
+  "blank_token_id": 4096,
   "final_dropout": 0.1,
   "hidden_act": "gelu",
   "hidden_dropout": 0.1,
   "hidden_size": 768,
   "initializer_range": 0.02,
   "input_dim": 80,
-  "intermediate_size": 3072,
+  "intermediate_size": 2048,
   "joiner_activation": "relu",
   "layer_norm_eps": 1e-05,
-  "layerdrop": 0.1,
   "left_context_length": 30,
   "lstm_dropout": 0.3,
   "lstm_hidden_dim": 512,
   "lstm_layer_norm": 0.001,
   "lstm_layer_norm_epsilon": 0.001,
-  "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": "emformer",
   "num_attention_heads": 8,
-  "num_codevector_groups": 2,
-  "num_codevectors_per_group": 320,
-  "num_conv_pos_embedding_groups": 16,
-  "num_conv_pos_embeddings": 128,
-  "num_feat_extract_layers": 7,
   "num_hidden_layers": 20,
   "num_lstm_layers": 3,
-  "num_negatives": 100,
   "output_dim": 1024,
-  "pad_token_id": 0,
-  "proj_codevector_dim": 256,
   "right_context_length": 4,
   "segment_length": 16,
   "symbol_embedding_dim": 512,
   "time_reduction_input_dim": 128,
   "time_reduction_stride": 4,
   "torch_dtype": "float32",
-  "transformer_ffn_dim": 2048,
   "transformers_version": "4.20.0.dev0",
   "vocab_size": 4097
 }

preprocessor_config.json

@@ -170,6 +170,7 @@
   "n_mels": 80,
   "padding_side": "right",
   "padding_value": 0.0,
+  "processor_class": "EmformerProcessor",
   "return_attention_mask": true,
   "sampling_rate": 16000
 }

tokenizer_config.json

@@ -1 +1 @@
-{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "sp_model_kwargs": {}, "tokenizer_class": "EmformerTokenizer"}
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "sp_model_kwargs": {}, "tokenizer_class": "EmformerTokenizer", "processor_class": "EmformerProcessor"}