refactor partial reinit script as a function. Allow user to select layers to reinit in finutunning

config.json

@@ -29,6 +29,8 @@
         "url": "tcp:\/\/localhost:54321"
     },
 
+    "reinit_layers": ["model.decoder.attention_layer"],
+
     "model": "Tacotron2",   // one of the model in models/    
     "grad_clip": 0.02,      // upper limit for gradients for clipping.
     "epochs": 1000,         // total number of epochs to train.

config_cluster.json

@@ -29,6 +29,8 @@
         "url": "tcp:\/\/localhost:54321"
     },
 
+    "reinit_layers": ["model.decoder.attention_layer"],
+
     "model": "Tacotron2",   // one of the model in models/    
     "grad_clip": 1,      // upper limit for gradients for clipping.
     "epochs": 1000,         // total number of epochs to train.

train.py

@@ -22,7 +22,8 @@ from utils.generic_utils import (NoamLR, check_update, count_parameters,
                                  create_experiment_folder, get_commit_hash,
                                  load_config, lr_decay,
                                  remove_experiment_folder, save_best_model,
-                                 save_checkpoint, sequence_mask, weight_decay)
+                                 save_checkpoint, sequence_mask, weight_decay,
+                                 set_init_dict)
 from utils.logger import Logger
 from utils.synthesis import synthesis
 from utils.text.symbols import phonemes, symbols
@@ -396,24 +397,9 @@ def main(args):
             print(" > Partial model initialization.")
             partial_init_flag = True
             model_dict = model.state_dict()
-            # Partial initialization: if there is a mismatch with new and old layer, it is skipped.
+            model_dict = set_init_dict(model_dict, checkpoint, c)
-            # 1. filter out unnecessary keys
-            pretrained_dict = {
-                k: v
-                for k, v in checkpoint['model'].items() if k in model_dict
-            }
-            # 2. filter out different size layers
-            pretrained_dict = {
-                k: v
-                for k, v in pretrained_dict.items()
-                if v.numel() == model_dict[k].numel()
-            }
-            # 3. overwrite entries in the existing state dict
-            model_dict.update(pretrained_dict)
-            # 4. load the new state dict
             model.load_state_dict(model_dict)
-            print(" | > {} / {} layers are initialized".format(
+            del model_dict
-                len(pretrained_dict), len(model_dict)))
         if use_cuda:
             model = model.cuda()
             criterion.cuda()

utils/generic_utils.py

@@ -195,3 +195,30 @@ def sequence_mask(sequence_length, max_len=None):
                          .expand_as(seq_range_expand))
     # B x T_max
     return seq_range_expand < seq_length_expand
+
+
+def set_init_dict(model_dict, checkpoint, c):
+    # Partial initialization: if there is a mismatch with new and old layer, it is skipped.
+    # 1. filter out unnecessary keys
+    pretrained_dict = {
+        k: v
+        for k, v in checkpoint['model'].items() if k in model_dict
+    }
+    # 2. filter out different size layers
+    pretrained_dict = {
+        k: v
+        for k, v in pretrained_dict.items()
+        if v.numel() == model_dict[k].numel()
+    }
+    # 3. skip reinit layers
+    if c.reinit_layers is not None:
+        for reinit_layer_name in c.reinit_layers:
+            pretrained_dict = {
+                k: v
+                for k, v in pretrained_dict.items()
+                if reinit_layer_name not in k
+            }
+    # 4. overwrite entries in the existing state dict
+    model_dict.update(pretrained_dict)
+    print(" | > {} / {} layers are initialized".format(len(pretrained_dict), len(model_dict)))
+    return model_dict