comments

Signed-off-by: sam1373 <[email protected]>

examples/asr/conf/citrinet/citrinet_384.yaml

@@ -54,6 +54,7 @@ model:
     se: true
     se_context_size: -1
     kernel_size_factor: 1.0
+    enc_final: 640
 
   tokenizer:
     dir: ???  # path to directory which contains either tokenizer.model (bpe) or vocab.txt (for wpe)
@@ -356,7 +357,7 @@ model:
         se_context_size: ${model.model_defaults.se_context_size}
         kernel_size_factor: ${model.model_defaults.kernel_size_factor}
 
-      - filters: &enc_final 640
+      - filters: ${model.model_defaults.enc_final}
         repeat: 1
         kernel: [41]
         stride: [1]
@@ -371,7 +372,7 @@ model:
 
   decoder:
     _target_: nemo.collections.asr.modules.ConvASRDecoder
-    feat_in: *enc_final
+    feat_in: ${model.model_defaults.enc_final}
     num_classes: -1  # filled with vocabulary size from tokenizer at runtime
     vocabulary: []  # filled with vocabulary from tokenizer at runtime
 

examples/asr/conf/citrinet/citrinet_512.yaml

@@ -53,6 +53,7 @@ model:
     se: true
     se_context_size: -1
     kernel_size_factor: 1.0
+    enc_final: 640
 
   tokenizer:
     dir: ???  # path to directory which contains either tokenizer.model (bpe) or vocab.txt (for wpe)
@@ -355,7 +356,7 @@ model:
         se_context_size: ${model.model_defaults.se_context_size}
         kernel_size_factor: ${model.model_defaults.kernel_size_factor}
 
-      - filters: &enc_final 640
+      - filters: ${model.model_defaults.enc_final}
         repeat: 1
         kernel: [41]
         stride: [1]
@@ -370,7 +371,7 @@ model:
 
   decoder:
     _target_: nemo.collections.asr.modules.ConvASRDecoder
-    feat_in: *enc_final
+    feat_in: ${model.model_defaults.enc_final}
     num_classes: -1  # filled with vocabulary size from tokenizer at runtime
     vocabulary: []  # filled with vocabulary from tokenizer at runtime
 

examples/asr/conf/ssl/citrinet/citrinet_ssl_1024.yaml

@@ -26,6 +26,7 @@ model:
     shuffle: true
     is_tarred: false
     tarred_audio_filepaths: null
+    shuffle_n: 2048
     use_start_end_token: false
     # bucketing params
     bucketing_strategy: "synced_randomized"

examples/asr/conf/ssl/conformer/conformer_ssl.yaml

@@ -33,11 +33,18 @@ model:
     batch_size: 16 # you may increase batch_size if your memory allows
     shuffle: true
     num_workers: 8
-    pin_memory: true
-    use_start_end_token: false
+    pin_memory: false
+    use_start_end_token: true
     trim_silence: false
-    max_duration: 35.0 # it is set for LibriSpeech, you may need to update it for your dataset
+    max_duration: 16.7 # it is set for LibriSpeech, you may need to update it for your dataset
     min_duration: 3.0
+    # tarred datasets
+    is_tarred: false
+    tarred_audio_filepaths: null
+    shuffle_n: 2048
+    # bucketing params
+    bucketing_strategy: "synced_randomized"
+    bucketing_batch_size: null
 
   validation_ds:
     manifest_filepath: ???
@@ -98,6 +105,7 @@ model:
 
     # Convolution module's params
     conv_kernel_size: 31
+    conv_norm_type: 'batch_norm' # batch_norm or layer_norm
 
     ### regularization
     dropout: 0.1 # The dropout used in most of the Conformer Modules
@@ -167,7 +175,7 @@ exp_manager:
     # in case of multiple validation sets, first one is used
     monitor: "val_loss"
     mode: "min"
-    save_top_k: 1
+    save_top_k: 5
 
   # you need to set these two to True to continue the training
   resume_if_exists: false

examples/asr/conf/ssl/contextnet/contextnet_ssl.yaml

@@ -26,12 +26,17 @@ model:
     trim_silence: false
     max_duration: 16.7
     shuffle: true
-    is_tarred: false
-    tarred_audio_filepaths: null
-    tarred_shard_strategy: "scatter"
     use_start_end_token: false
     num_workers: 16
     pin_memory: true
+    # tarred datasets
+    is_tarred: false
+    tarred_audio_filepaths: null
+    tarred_shard_strategy: "scatter"
+    shuffle_n: 2048
+    # bucketing params
+    bucketing_strategy: "synced_randomized"
+    bucketing_batch_size: null
 
   validation_ds:
     manifest_filepath: ???

nemo/core/classes/modelPT.py

@@ -949,7 +949,7 @@ def maybe_init_from_pretrained_checkpoint(self, cfg: OmegaConf, map_location: st
                     self.load_state_dict(restored_model.state_dict(), strict=False)
                     logging.info(f'Model checkpoint restored from nemo file with path : `{model_path}`')
                     del restored_model
-                else:
+                elif isinstance(cfg.init_from_nemo_model, dict):
                     model_load_dict = cfg.init_from_nemo_model
                     for model_load_cfg in model_load_dict.values():
                         model_path = model_load_cfg.path
@@ -966,6 +966,8 @@ def maybe_init_from_pretrained_checkpoint(self, cfg: OmegaConf, map_location: st
                         )
 
                         del restored_model
+                else:
+                    raise TypeError("Invalid type: init_from_nemo_model is not a string or a dict!")
 
         if 'init_from_pretrained_model' in cfg and cfg.init_from_pretrained_model is not None:
             with open_dict(cfg):
@@ -996,7 +998,7 @@ def maybe_init_from_pretrained_checkpoint(self, cfg: OmegaConf, map_location: st
                     logging.info(f'Model checkpoint restored from pretrained chackpoint with name : `{model_name}`')
 
                     del restored_model
-                else:
+                elif isinstance(cfg.init_from_pretrained_model, dict):
                     model_load_dict = cfg.init_from_pretrained_model
                     for model_load_cfg in model_load_dict.values():
                         model_name = model_load_cfg.name
@@ -1016,6 +1018,8 @@ def maybe_init_from_pretrained_checkpoint(self, cfg: OmegaConf, map_location: st
                         )
 
                         del restored_model
+                else:
+                    raise TypeError("Invalid type: init_from_pretrained_model is not a string or a dict!")
 
         if 'init_from_ptl_ckpt' in cfg and cfg.init_from_ptl_ckpt is not None:
             with open_dict(cfg):
@@ -1031,7 +1035,7 @@ def maybe_init_from_pretrained_checkpoint(self, cfg: OmegaConf, map_location: st
                     )
 
                     del ckpt
-                else:
+                elif isinstance(cfg.init_from_ptl_ckpt, dict):
                     model_load_dict = cfg.init_from_ptl_ckpt
                     for model_load_cfg in model_load_dict.values():
                         ckpt_path = model_load_cfg.path
@@ -1046,6 +1050,8 @@ def maybe_init_from_pretrained_checkpoint(self, cfg: OmegaConf, map_location: st
                         )
 
                         del ckpt
+                else:
+                    raise TypeError("Invalid type: init_from_ptl_ckpt is not a string or a dict!")
 
     def teardown(self, stage: str):
         """

tutorials/asr/README.md

@@ -29,7 +29,7 @@ In this repository, you will find several tutorials discussing what is Automatic
 
 10) `ASR_with_Transducers`: In this tutorial, we take a deep dive into Transducer based ASR models, discussing the similarity of setup and config to CTC models and then train a small ContextNet model on the AN4 dataset. We then discuss how to change the decoding strategy of a trained Transducer from greedy search to beam search. Finally, we wrap up this tutorial by extraining the alignment matrix from a trained Transducer model. 
 
-11) `Self_Supervised_Pre_Training`: It can often be difficult to obtain labeled data for ASR training. In this tutorial, we demonstrate how to pre-train a small Citrinet model in an unsupervised manner, and then fine-tune with CTC loss.
+11) `Self_Supervised_Pre_Training`: It can often be difficult to obtain labeled data for ASR training. In this tutorial, we demonstrate how to pre-train a speech model in an unsupervised manner, and then fine-tune with CTC loss.
 
 ----------------