Spaces:
Runtime error
Runtime error
adriansd12
commited on
Commit
β’
c6e8a33
1
Parent(s):
4c628ba
init. commit
Browse files- app.py +36 -0
- module/bible_index.py +53 -0
- requirements.txt +2 -0
app.py
ADDED
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from module.bible_index import BibleIndex
|
3 |
+
|
4 |
+
|
5 |
+
def query_index(query, testament, top_n):
|
6 |
+
_index = BibleIndex(testament)
|
7 |
+
items = _index.query(query, top_n=top_n)
|
8 |
+
|
9 |
+
item_list = f"<h2>{query}</h2>"
|
10 |
+
item_list += "<ul>"
|
11 |
+
for item in items:
|
12 |
+
item_list += f"<h3>{item.get('src')}</h3>"
|
13 |
+
item_list += f"<li>{item.get('text')}</li>"
|
14 |
+
item_list += "</ul>"
|
15 |
+
return item_list
|
16 |
+
|
17 |
+
|
18 |
+
demo = gr.Interface(
|
19 |
+
query_index,
|
20 |
+
[
|
21 |
+
gr.Textbox(label="Query text"),
|
22 |
+
gr.Radio(["all", "old", "new"], label="Section of the Bible"),
|
23 |
+
gr.Slider(0, 10, step=1, label="Top N results"),
|
24 |
+
],
|
25 |
+
outputs="html",
|
26 |
+
examples=[
|
27 |
+
["What is love", "new", 5],
|
28 |
+
["How old was Adam?", "old", 3],
|
29 |
+
["Who is God?", "all", 7],
|
30 |
+
],
|
31 |
+
title="Bible Search Index",
|
32 |
+
description="""
|
33 |
+
A search index for The Bible using *sentence_transformer*.
|
34 |
+
""",
|
35 |
+
)
|
36 |
+
demo.launch()
|
module/bible_index.py
ADDED
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import numpy as np
|
2 |
+
from sentence_transformers import SentenceTransformer, util
|
3 |
+
|
4 |
+
|
5 |
+
class BibleIndex:
|
6 |
+
def __init__(self, testament: str = "all") -> None:
|
7 |
+
self.model = SentenceTransformer(
|
8 |
+
"sentence-transformers/msmarco-bert-base-dot-v5"
|
9 |
+
)
|
10 |
+
|
11 |
+
match testament:
|
12 |
+
case "all" | "old" | "new":
|
13 |
+
self.testament = testament
|
14 |
+
case _:
|
15 |
+
print("error:")
|
16 |
+
|
17 |
+
self.load_emb()
|
18 |
+
self.load_text()
|
19 |
+
|
20 |
+
def load_emb(self) -> None:
|
21 |
+
self.emb = np.load(f"data/embeddings/{self.testament}_esv_embeddings.npy")
|
22 |
+
|
23 |
+
def load_text(self) -> None:
|
24 |
+
text_path = f"data/text/{self.testament}_testament_esv.txt"
|
25 |
+
|
26 |
+
with open(text_path, "r") as f:
|
27 |
+
self.text = f.readlines()[1:]
|
28 |
+
|
29 |
+
def query(self, query: str = "", top_n: int = 10):
|
30 |
+
query_emb = self.model.encode(query)
|
31 |
+
scores = util.dot_score(query_emb, self.emb)[0].cpu().tolist()
|
32 |
+
|
33 |
+
# Combine docs & scores
|
34 |
+
doc_score_pairs = list(zip(self.text, scores))
|
35 |
+
|
36 |
+
# Sort by decreasing score
|
37 |
+
doc_score_pairs = sorted(doc_score_pairs, key=lambda x: x[1], reverse=True)
|
38 |
+
|
39 |
+
# Output passages & scores
|
40 |
+
print("Query:", query)
|
41 |
+
results = []
|
42 |
+
for doc, score in doc_score_pairs[:top_n]:
|
43 |
+
text_split = doc.split(",")
|
44 |
+
results.append(
|
45 |
+
{
|
46 |
+
"src": f"{text_split[0]} {text_split[1]}:{text_split[2]}",
|
47 |
+
"text": ",".join(text_split[3:])
|
48 |
+
.replace("\xa0", "")
|
49 |
+
.replace("\n", ""),
|
50 |
+
"score": score,
|
51 |
+
}
|
52 |
+
)
|
53 |
+
return results
|
requirements.txt
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
numpy==1.24.2
|
2 |
+
sentence-transformers==2.2.2
|