Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -39,27 +39,27 @@ st.title('Harry Potter and the Extractive Question Answering Model')
|
|
39 |
# Type in HP-related query here
|
40 |
query = st.text_area("Hello my dears! What is your question? Be patient please, I am not a Ravenclaw!")
|
41 |
|
42 |
-
# Perform sentence embedding on query and sentence groups
|
43 |
-
model_embed_name = 'sentence-transformers/msmarco-distilbert-dot-v5'
|
44 |
-
|
45 |
-
model_embed = SentenceTransformer(model_embed_name)
|
46 |
-
doc_emb = model_embed.encode(paragraphs)
|
47 |
-
query_emb = model_embed.encode(query)
|
48 |
-
|
49 |
-
#Compute dot score between query and all document embeddings
|
50 |
-
scores = util.cos_sim(query_emb, doc_emb)[0].cpu().tolist()
|
51 |
-
|
52 |
-
#Combine docs & scores
|
53 |
-
doc_score_pairs = list(zip(paragraphs, scores))
|
54 |
-
|
55 |
-
#Sort by decreasing score and get only 3 most similar groups
|
56 |
-
doc_score_pairs = sorted(doc_score_pairs, key=lambda x: x[1],
|
57 |
-
reverse=True)[:3]
|
58 |
-
|
59 |
-
# Join these similar groups to form the context
|
60 |
-
context = "".join(x[0] for x in doc_score_pairs)
|
61 |
-
|
62 |
if st.button('Ask'):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
63 |
# Perform the querying
|
64 |
QA_input = {'question': query, 'context': context}
|
65 |
res = pipe(QA_input)
|
|
|
39 |
# Type in HP-related query here
|
40 |
query = st.text_area("Hello my dears! What is your question? Be patient please, I am not a Ravenclaw!")
|
41 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
if st.button('Ask'):
|
43 |
+
# Perform sentence embedding on query and sentence groups
|
44 |
+
model_embed_name = 'sentence-transformers/msmarco-distilbert-dot-v5'
|
45 |
+
|
46 |
+
model_embed = SentenceTransformer(model_embed_name)
|
47 |
+
doc_emb = model_embed.encode(paragraphs)
|
48 |
+
query_emb = model_embed.encode(query)
|
49 |
+
|
50 |
+
#Compute dot score between query and all document embeddings
|
51 |
+
scores = util.cos_sim(query_emb, doc_emb)[0].cpu().tolist()
|
52 |
+
|
53 |
+
#Combine docs & scores
|
54 |
+
doc_score_pairs = list(zip(paragraphs, scores))
|
55 |
+
|
56 |
+
#Sort by decreasing score and get only 3 most similar groups
|
57 |
+
doc_score_pairs = sorted(doc_score_pairs, key=lambda x: x[1],
|
58 |
+
reverse=True)[:3]
|
59 |
+
|
60 |
+
# Join these similar groups to form the context
|
61 |
+
context = "".join(x[0] for x in doc_score_pairs)
|
62 |
+
|
63 |
# Perform the querying
|
64 |
QA_input = {'question': query, 'context': context}
|
65 |
res = pipe(QA_input)
|