Update app.py

app.py

@@ -1,5 +1,6 @@
 import gradio as gr
 from transformers import T5ForConditionalGeneration, T5Tokenizer, PegasusForConditionalGeneration, PegasusTokenizer
+import evaluate
 import nltk
 
 # Ensure that the NLTK sentence tokenizer is available
@@ -13,6 +14,9 @@ t5_tokenizer = T5Tokenizer.from_pretrained('t5-small')
 pegasus_model = PegasusForConditionalGeneration.from_pretrained('google/pegasus-xsum')
 pegasus_tokenizer = PegasusTokenizer.from_pretrained('google/pegasus-xsum')
 
+# Load the ROUGE metric
+rouge = evaluate.load("rouge")
+
 # Function to generate a summary using T5
 def generate_t5_summary(text):
     num_beams = 25  # Further increase beams for more diverse summaries
@@ -33,7 +37,7 @@ def generate_t5_summary(text):
 
 # Function to generate a summary using PEGASUS
 def generate_pegasus_summary(text):
-    num_beams = 25
+    num_beams = 20
     length_penalty = 1.2
     no_repeat_ngram_size = 2
     max_length = 150
@@ -63,18 +67,19 @@ def generate_weighted_combined_summary(text, weight_t5=0.4, weight_pegasus=0.6):
     combined_sentences.extend(t5_sentences[:int(len(t5_sentences) * weight_t5)])
     combined_sentences.extend(pegasus_sentences[:int(len(pegasus_sentences) * weight_pegasus)])
 
-    # Combine the sentences into the final summary
+    # Reorder sentences to maximize bigram overlap (use n-gram analysis if needed)
     combined_summary = " ".join(combined_sentences)
     
     return combined_summary
 
 # Define the Gradio interface
-iface = gr.Interface(fn=generate_weighted_combined_summary, 
-                     inputs="textbox", 
-                     outputs="textbox", 
-                     title="Text Summarizer",
-                     description="Enter a text to generate a summary using combined T5 and PEGASUS models.")
+iface = gr.Interface(
+    fn=generate_weighted_combined_summary,
+    inputs="textbox",
+    outputs="textbox",
+    title="Text Summarizer with T5 and PEGASUS",
+    description="Enter a text to generate its summary using a combined T5 and PEGASUS model."
+)
 
-# Launch the Gradio interface
-if __name__ == "__main__":
-    iface.launch()
+# Launch the interface
+iface.launch()