Update app.py

app.py

@@ -15,11 +15,8 @@ torch_device = 'cuda' if torch.cuda.is_available() else 'cpu'
 tokenizer3 = PegasusTokenizer.from_pretrained(model_name)
 model3 = PegasusForConditionalGeneration.from_pretrained(model_name).to(torch_device)
 
-prev_context = "" # to store the previous context
 
 def qa_paraphrase(text_input, question):
-    global prev_context
-    text_input = prev_context + " " + text_input # combine with previous context
     prediction = classifier(
         context=text_input,
         question=question,
@@ -38,8 +35,7 @@ def qa_paraphrase(text_input, question):
     batch = tokenizer3([sentence],truncation=True,padding='longest',max_length=60, return_tensors="pt").to(torch_device)
     translated = model3.generate(**batch,max_length=60,num_beams=10, num_return_sequences=1, temperature=1.5)
     paraphrase = tokenizer3.batch_decode(translated, skip_special_tokens=True)[0]
-    prev_context = text_input # update previous context
-    return f"Q: {question}\nA: {answer}\nParaphrased Sentence: {paraphrase}"
+    return f"Answer: {answer}\nLong Form Answer: {paraphrase}"
 
 
 iface = gr.Interface(
@@ -49,8 +45,8 @@ iface = gr.Interface(
         gr.inputs.Textbox(label="Question")
     ],
     outputs=gr.outputs.Textbox(label="Output"),
-    title="Question Answering and Paraphrasing",
-    description="Answer a given question from a given text and paraphrase the sentence containing the answer."
+    title="Long Form Question Answering",
+    description="mimics long form question answering by extracting the sentence containing the answer and paraphrasing it"
 )
 
-iface.launch() 
+iface.launch()