Update app.py

app.py

@@ -52,7 +52,21 @@ def predict(query, context):
         cust_answer_sentence = cust_answer_sentence[0] 
     else:
         cust_answer_sentence = "Failed matching sentence (answer may be split in multiple sentences)"
-        
+    
+    #Custom3
+    cust_model_name_3 = "consciousAI/question-answering-roberta-base-s-v2"
+    cust_question_answerer_3 = pipeline('question-answering', model=cust_model_name_3, tokenizer=cust_model_name_3, device=device)
+    
+    cust_output_3 = cust_question_answerer_3(question=query, context=context)
+    cust_answer_3 = cust_output_3['answer']
+    cust_answer_span_3 = "[" + str(cust_output_3['start']) + "," + str(cust_output_3['end']) + "]" 
+    cust_confidence_3 = cust_output_3['score']
+    cust_answer_sentence_3 = [_sent for _sent in sent_tokenize(context) if cust_answer_3 in _sent]
+    if len(cust_answer_sentence_3) > 0:
+        cust_answer_sentence_3 = cust_answer_sentence_3[0] 
+    else:
+        cust_answer_sentence_3 = "Failed matching sentence (answer may be split in multiple sentences)"
+               
     #Custom2
     cust_answer_2 = _generate(query, context, model="consciousAI/question-answering-generative-t5-v1-base-s-q-c", device=device)
     cust_answer_sentence_2 = [_sent for _sent in sent_tokenize(context) if cust_answer_2 in _sent]
@@ -62,7 +76,7 @@ def predict(query, context):
         cust_answer_sentence_2 = "Failed matching sentence (answer may be split in multiple sentences)"
     cust_answer_span_2 = re.search(cust_answer_2, contextDefault).span()
     
-    return cust_answer, cust_answer_sentence, cust_answer_span, cust_confidence, cust_answer_2, cust_answer_sentence_2, cust_answer_span_2
+    return cust_answer, cust_answer_sentence, cust_answer_span, cust_confidence, cust_answer_2, cust_answer_sentence_2, cust_answer_span_2, cust_answer_3, cust_answer_sentence_3, cust_answer_span_3, cust_confidence_3
     
 with gr.Blocks() as demo:
     gr.Markdown(value="# Question Answering Encoders vs Generative\n [Question Answering Leveraging Encoders](https://huggingface.co/anshoomehra/question-answering-roberta-base-s)\n\n[Generative Question Answering](https://huggingface.co/anshoomehra/question-answering-generative-t5-v1-base-s-q-c)")
@@ -78,8 +92,13 @@ with gr.Blocks() as demo:
             with gr.Column(variant='compact'):
                 _predictionM6 = gr.Textbox(label="question-answering-roberta-base-s: Answer Sentence")
                 _predictionM5 = gr.Textbox(label="question-answering-roberta-base-s: Answer")
-                _predictionM7 = gr.Textbox(label="question-answering-roberta-base-s:Cisco Q&A Answer Span")
+                _predictionM7 = gr.Textbox(label="question-answering-roberta-base-s: Q&A Answer Span")
                 _predictionM8 = gr.Textbox(label="question-answering-roberta-base-s: Answer Confidence")
+            with gr.Column(variant='compact'):
+                _predictionM12 = gr.Textbox(label="question-answering-roberta-base-s-v2: Answer Sentence")
+                _predictionM13 = gr.Textbox(label="question-answering-roberta-base-s-v2: Answer")
+                _predictionM14 = gr.Textbox(label="question-answering-roberta-base-s-v2: Q&A Answer Span")
+                _predictionM15 = gr.Textbox(label="question-answering-roberta-base-s-v2: Answer Confidence")
             with gr.Column(variant='compact'):   
                 _predictionM10 = gr.Textbox(label="question-answering-generative-t5-v1-base-s-q-c: Sentence")
                 _predictionM9 = gr.Textbox(label="question-answering-generative-t5-v1-base-s-q-c: Answer")
@@ -90,7 +109,7 @@ with gr.Blocks() as demo:
         gen_btn = gr.Button("Generate Answers")
         gen_btn.click(fn=predict,
                       inputs=[query, context],
-                      outputs=[_predictionM5, _predictionM6, _predictionM7, _predictionM8, _predictionM9, _predictionM10, _predictionM11]
+                      outputs=[_predictionM5, _predictionM6, _predictionM7, _predictionM8, _predictionM9, _predictionM10, _predictionM11, _predictionM12, _predictionM13, _predictionM14, _predictionM15]
                       )
 
 demo.launch(show_error=True)