Reorganize ALBERT conversion script

transformers/modeling_albert.py

@@ -68,14 +68,36 @@ def load_tf_weights_in_albert(model, config, tf_checkpoint_path):
     
     for name, array in zip(names, arrays):
         original_name = name
+
+        # If saved from the TF HUB module
+        name = name.replace("module/", "")
+
+        # Renaming and simplifying
         name = name.replace("ffn_1", "ffn")
-        name = name.replace("/bert/", "/albert/")
-        name = name.replace("ffn/intermediate/output", "ffn_output")
+        name = name.replace("bert/", "albert/")
         name = name.replace("attention_1", "attention")   
-        name = name.replace("cls/predictions", "predictions")
         name = name.replace("transform/", "")
         name = name.replace("LayerNorm_1", "full_layer_layer_norm")    
-        name = name.replace("LayerNorm", "attention/LayerNorm")    
+        name = name.replace("LayerNorm", "attention/LayerNorm")   
+        name = name.replace("transformer/", "")
+
+        # The feed forward layer had an 'intermediate' step which has been abstracted away
+        name = name.replace("intermediate/dense/", "")
+        name = name.replace("ffn/intermediate/output/dense/", "ffn_output/")
+
+        # ALBERT attention was split between self and output which have been abstracted away
+        name = name.replace("/output/", "/")
+        name = name.replace("/self/", "/")
+
+        # The pooler is a linear layer
+        name = name.replace("pooler/dense", "pooler")
+
+        # The classifier was simplified to predictions from cls/predictions
+        name = name.replace("cls/predictions", "predictions")
+        name = name.replace("predictions/attention", "predictions")
+
+        # Naming was changed to be more explicit
+        name = name.replace("embeddings/attention", "embeddings")    
         name = name.replace("inner_group_", "albert_layers/") 
         name = name.replace("group_", "albert_layer_groups/")    
         name = name.split('/')