tensorflow
diff --git a/‎Examples/BERT-CoLA/main.swift‎
Lines changed: 1 addition & 1 deletion b/‎Examples/BERT-CoLA/main.swift‎
Lines changed: 1 addition & 1 deletion
diff --git a/‎Models/Text/Attention.swift‎ ‎Models/Text/BERT/Attention.swift‎Models/Text/Attention.swift renamed to Models/Text/BERT/Attention.swift b/‎Models/Text/Attention.swift‎ ‎Models/Text/BERT/Attention.swift‎Models/Text/Attention.swift renamed to Models/Text/BERT/Attention.swift
diff --git a/‎Models/Text/BERT.swift‎ ‎Models/Text/BERT/BERT.swift‎Models/Text/BERT.swift renamed to Models/Text/BERT/BERT.swift b/‎Models/Text/BERT.swift‎ ‎Models/Text/BERT/BERT.swift‎Models/Text/BERT.swift renamed to Models/Text/BERT/BERT.swift
diff --git a/‎Models/Text/BERTClassifier.swift‎ ‎Models/Text/BERT/BERTClassifier.swift‎Models/Text/BERTClassifier.swift renamed to Models/Text/BERT/BERTClassifier.swift b/‎Models/Text/BERTClassifier.swift‎ ‎Models/Text/BERT/BERTClassifier.swift‎Models/Text/BERTClassifier.swift renamed to Models/Text/BERT/BERTClassifier.swift
diff --git a/‎Models/Text/TransformerBERT.swift‎ ‎Models/Text/BERT/TransformerBERT.swift‎Models/Text/TransformerBERT.swift renamed to Models/Text/BERT/TransformerBERT.swift b/‎Models/Text/TransformerBERT.swift‎ ‎Models/Text/BERT/TransformerBERT.swift‎Models/Text/TransformerBERT.swift renamed to Models/Text/BERT/TransformerBERT.swift
diff --git a/‎Models/Text/Utilities.swift‎ ‎Models/Text/BERT/Utilities.swift‎Models/Text/Utilities.swift renamed to Models/Text/BERT/Utilities.swift b/‎Models/Text/Utilities.swift‎ ‎Models/Text/BERT/Utilities.swift‎Models/Text/Utilities.swift renamed to Models/Text/BERT/Utilities.swift
diff --git a/‎Models/Text/CMakeLists.txt‎
Lines changed: 5 additions & 6 deletions b/‎Models/Text/CMakeLists.txt‎
Lines changed: 5 additions & 6 deletions
diff --git a/‎Models/Text/WeightDecayedAdam.swift‎
Lines changed: 0 additions & 55 deletions b/‎Models/Text/WeightDecayedAdam.swift‎
Lines changed: 0 additions & 55 deletions
@@ -87,7 +87,7 @@ let useBiasCorrection = true
 var optimizer = x10_optimizers_optimizer.GeneralOptimizer(
   for: bertClassifier,
   TensorVisitorPlan(bertClassifier.differentiableVectorView),
-  defaultOptimizer: makeWeightDecayedAdam(
+  defaultOptimizer: makeAdam(
     learningRate: peakLearningRate,
     beta1: beta1,
     beta2: beta2
 
@@ -1,16 +1,15 @@
 add_library(TextModels
-  Attention.swift
-  BERT.swift
-  BERTClassifier.swift
+  BERT/Attention.swift
+  BERT/BERT.swift
+  BERT/BERTClassifier.swift
   BERT/BERTCheckpointReader.swift
+  BERT/TransformerBERT.swift
+  BERT/Utilities.swift
   GPT2/CheckpointWriter.swift
   GPT2/GPT2.swift
   GPT2/TransformerLM.swift
   GPT2/Operators.swift
   GPT2/PythonCheckpointReader.swift
-  TransformerBERT.swift
-  Utilities.swift
-  WeightDecayedAdam.swift
   WordSeg/Lattice.swift
   WordSeg/Model.swift
   WordSeg/SemiRing.swift)