Formatted generate_text output

app.py

@@ -26,6 +26,7 @@ from tensorflow import keras
 import keras_nlp
 #from google.colab import drive
 import time
+import re
 
 os.environ["KERAS_BACKEND"] = "tensorflow"  # or "tensorflow" or "torch"
 
@@ -46,8 +47,31 @@ gpt2_lm.load_weights(checkpoint_path)
 
 """# Chargement et configuration de gradio"""
 
+"""
 def generate_text(prompt):
     return gpt2_lm.generate(prompt, max_length=100)
+"""
+
+# Expérimental
+def format_text(text, to_remove):
+    # Function to format generated text
+    # - removes the prompt from the answer
+    # - removes unnecessary space chars before punctuation
+    # - capitalize the first letter of each sentence
+    text = text.replace(to_remove, '')
+    text = re.sub(r' +,', ',', text) # comma
+    text = re.sub(r' +\.', '.', text) # full stop
+    text = '. '.join(sentence.strip().capitalize() for sentence in text.split('.'))
+
+    return text
+
+# Expérimental
+def generate_text(prompt):
+  output = gpt2_lm.generate(prompt, max_length=150)
+  formatted_output = format_text(output, prompt)
+  print(f"DEBUG - GPT-2 Output : {output}")
+  print(f"DEBUG - Formatted GPT-2 Output : {formatted_output}")
+  return formatted_output
 
 # CSS styles
 css = """