Rename clqna.py to demo.py
Browse files
clqna.py
DELETED
@@ -1,35 +0,0 @@
|
|
1 |
-
import streamlit as st
|
2 |
-
|
3 |
-
from transformers import pipeline
|
4 |
-
from pinecone import Pinecone, ServerlessSpec
|
5 |
-
from sentence_transformers import SentenceTransformer, util
|
6 |
-
|
7 |
-
|
8 |
-
bi_encoder = SentenceTransformer('msmarco-distilbert-base-v4')
|
9 |
-
bi_encoder.max_seq_length = 256 # Truncate long documents to 256 tokens
|
10 |
-
|
11 |
-
# Store the index as a variable
|
12 |
-
INDEX_NAME = 'cl-search-idx'
|
13 |
-
NAMESPACE = 'webpages'
|
14 |
-
|
15 |
-
index = pc.Index(name=INDEX_NAME)
|
16 |
-
|
17 |
-
def query_from_pinecone(index, question_embedding, top_k=3):
|
18 |
-
# get embedding from THE SAME embedder as the documents
|
19 |
-
|
20 |
-
return index.query(
|
21 |
-
vector=question_embedding,
|
22 |
-
top_k=top_k,
|
23 |
-
namespace=NAMESPACE,
|
24 |
-
include_metadata=True # gets the metadata (dates, text, etc)
|
25 |
-
).get('matches')
|
26 |
-
|
27 |
-
|
28 |
-
QUESTION=st.text_area('Ask a question -e.g How to prepare for Verbal section for CAT?') ##' How to prepare for Verbal section ?'
|
29 |
-
question_embedding = bi_encoder.encode(QUESTION, convert_to_tensor=True)
|
30 |
-
resp= query_from_pinecone(question_embedding.tolist(), 3)
|
31 |
-
docresult= resp[0]['metadata']['text']
|
32 |
-
#+ '\n*************\n'+ resp[1]['metadata']['text'] + '\n*************\n'+ resp[2]['metadata']['text']
|
33 |
-
|
34 |
-
st.json(out)
|
35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
demo.py
ADDED
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import transformers
|
3 |
+
from transformers import pipeline
|
4 |
+
|
5 |
+
pipe= pipeline('sentiment-analysis')
|
6 |
+
|
7 |
+
text = st.text_area('Enter some text')
|
8 |
+
|
9 |
+
if text:
|
10 |
+
out= pipe(text)
|
11 |
+
st.json(out)
|