Create app.py

app.py

@@ -0,0 +1,63 @@
+import streamlit as st
+from sentence_transformers import SentenceTransformer, util
+from transformers import (AutoModelForQuestionAnswering,
+                          AutoTokenizer, pipeline)
+
+import pandas as pd
+import regex as re
+
+# Select model for question answering
+model_name = "deepset/roberta-base-squad2"
+
+# Load model & tokenizer
+model = AutoModelForQuestionAnswering.from_pretrained(model_name)
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+
+# Create pipeline
+pipe = pipeline('question-answering', model=model_name, tokenizer=model_name)
+
+# Load DFA Press Release dataset
+df = pd.read_csv('../data/dfa_pr_v4_cleaned.csv')
+
+# Group into 6 sentences-long parts
+partitions = df['article'].values.tolist()
+
+st.title('DFA Question Answering Model')
+
+# Type in HP-related query here
+query = st.text_area("Type in your question below:")
+
+if st.button('Searching for the answer...'):
+    # Perform sentence embedding on query and sentence groups
+    model_embed_name = 'sentence-transformers/msmarco-distilbert-dot-v5'
+    
+    model_embed = SentenceTransformer(model_embed_name)
+    doc_emb = model_embed.encode(partitions)
+    query_emb = model_embed.encode(query)
+    
+    #Compute dot score between query and all document embeddings
+    scores = util.dot_score(query_emb, doc_emb)[0].cpu().tolist()
+    
+    #Combine docs & scores
+    doc_score_pairs = list(zip(partitions, scores))
+    
+    #Sort by decreasing score and get only 3 most similar groups
+    doc_score_pairs = sorted(doc_score_pairs, key=lambda x: x[1],
+                         reverse=True)[:1]
+    
+    # Join these similar groups to form the context 
+    context = "".join(x[0] for x in doc_score_pairs)
+
+    # Perform the querying
+    QA_input = {'question': query, 'context': context}
+    res = pipe(QA_input)
+    
+    confidence = res.get('score')
+    if confidence > 0.8:
+        st.write(res.get('answer'))
+        st.write(context)
+    else:
+        out = "I am not sure."
+        st.write(out)
+        st.write(context)
+    #out = res.get('answer')