comments updated

README.md

@@ -12,7 +12,7 @@ probably proofread and complete it, then remove this comment. -->
 
 # Manirathinam21/DistilBert_SMSSpam_classifier
 
-This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on an unknown dataset.
+This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on an SMSSpam Detection dataset.
 It achieves the following results on the evaluation set:
 - Train Loss: 0.0114
 - Train Accuracy: 0.9962
@@ -26,7 +26,13 @@ label: a classification label, with possible values including
 
 ## Model description
 
-More information needed
+Tokenizer used is DistilBertTokenizerFast  with return_tensors='tf' parameter in tokenizer because building model in a tensorflow framework
+
+Model: TFDistilBertForSequenceClassification 
+ 
+Optimizer: Adam with learning rate=5e-5
+
+Loss: SparseCategoricalCrossentropy
 
 ## Intended uses & limitations
 
@@ -38,6 +44,10 @@ More information needed
 
 ## Training procedure
 
+After Tokenized, Encoded datasets are converted to Dataset Objects by using tf.data.Dataset.from_tensor_slices((dict(train_encoding),  train_y))
+
+This step is done to inject a dataset into TFModel in a specific TF format
+
 ### Training hyperparameters
 
 The following hyperparameters were used during training: