torch.no_grad vs. requires_grad

4 files changed, 35 insertions, 0 deletions

diff --git a/fine_tune/bert/tutorials/03_bert_input_embedding.py b/fine_tune/bert/tutorials/03_bert_input_embedding.py
new file mode 100644
index 0000000..41bdd29
--- /dev/null
+++ b/fine_tune/bert/tutorials/03_bert_input_embedding.py
@@ -0,0 +1,20 @@
+
+from transformers import BertTokenizer, BertModel
+from transformers.models.bert import BertModel
+import torch
+from torch import nn
+
+
+model_name = 'bert-base-uncased'
+
+tokenizer = BertTokenizer.from_pretrained(model_name)
+model = BertModel.from_pretrained(model_name, output_hidden_states=True)
+
+test_sent = 'this is a test sentence'
+
+model_input = tokenizer(test_sent, return_tensors='pt')
+
+
+model.eval()
+with torch.no_grad():
+    output = model(**model_input)
diff --git a/fine_tune/bert/tutorials/samples/BERT-embeddings-2.png b/fine_tune/bert/tutorials/samples/BERT-embeddings-2.png
new file mode 100644
index 0000000..9b7072e
--- /dev/null
+++ b/fine_tune/bert/tutorials/samples/BERT-embeddings-2.png
diff --git a/fine_tune/bert_parameters.py b/fine_tune/bert_parameters.py
index 98e2b15..bf6e8c1 100644
--- a/fine_tune/bert_parameters.py
+++ b/fine_tune/bert_parameters.py
@@ -9,6 +9,8 @@ model_name = 'bert-base-uncased'
 model = BertModel.from_pretrained(model_name)
 cls_model = BertForSequenceClassification.from_pretrained(model_name)
 
+
+
 total_params = 0
 total_learnable_params = 0
 total_embedding_params = 0
diff --git a/fine_tune/input_output.py b/fine_tune/input_output.py
new file mode 100644
index 0000000..684ded5
--- /dev/null
+++ b/fine_tune/input_output.py
@@ -0,0 +1,13 @@
+
+from transformers import BertModel, BertTokenizer
+
+model_name = 'bert-base-uncased'
+
+tokenizer = BertTokenizer.from_pretrained(model_name)
+model = BertModel.from_pretrained(model_name)
+
+raw_sentences = ['Tom likes cats', 'Liz likes dogs']
+
+inputs = tokenizer.encode_plus(raw_sentences[0], raw_sentences[1], return_tensors='pt')
+# inputs = tokenizer('Hello, my dog is cute', return_tensors='pt')
+model(**inputs)