Spaces:
Running
Running
My Duong
commited on
Commit
·
d594c97
1
Parent(s):
764c9b9
update app
Browse files- app.py +6 -0
- requirements.txt +2 -1
app.py
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
import os
|
|
|
2 |
import numpy as np
|
3 |
from pathlib import Path
|
4 |
import zipfile
|
@@ -9,6 +10,7 @@ from bs4 import BeautifulSoup
|
|
9 |
from sentence_transformers import SentenceTransformer
|
10 |
from torch.utils.data import DataLoader
|
11 |
from accelerate import Accelerator
|
|
|
12 |
|
13 |
|
14 |
# Wrapper for embedding
|
@@ -153,6 +155,10 @@ def save_to_chromadb(
|
|
153 |
# Persist changes to disk.
|
154 |
vector_db.persist()
|
155 |
print(f"Database saved successfully to {persist_directory}")
|
|
|
|
|
|
|
|
|
156 |
return vector_db
|
157 |
|
158 |
|
|
|
1 |
import os
|
2 |
+
import shutil
|
3 |
import numpy as np
|
4 |
from pathlib import Path
|
5 |
import zipfile
|
|
|
10 |
from sentence_transformers import SentenceTransformer
|
11 |
from torch.utils.data import DataLoader
|
12 |
from accelerate import Accelerator
|
13 |
+
from datasets import Dataset
|
14 |
|
15 |
|
16 |
# Wrapper for embedding
|
|
|
155 |
# Persist changes to disk.
|
156 |
vector_db.persist()
|
157 |
print(f"Database saved successfully to {persist_directory}")
|
158 |
+
|
159 |
+
shutil.make_archive("chroma_db", "zip", "./chroma_db")
|
160 |
+
print("Vector database archived as chroma_db.zip")
|
161 |
+
|
162 |
return vector_db
|
163 |
|
164 |
|
requirements.txt
CHANGED
@@ -19,4 +19,5 @@ chromadb
|
|
19 |
tiktoken
|
20 |
readability-lxml
|
21 |
gradio
|
22 |
-
langchain-chroma>=0.1.2
|
|
|
|
19 |
tiktoken
|
20 |
readability-lxml
|
21 |
gradio
|
22 |
+
langchain-chroma>=0.1.2
|
23 |
+
datasets
|