Spaces:
Sleeping
Sleeping
Update initialize.py
Browse files- initialize.py +35 -9
initialize.py
CHANGED
@@ -22,23 +22,49 @@ embedding_model = OpenAIEmbeddings(api_key=OPENAI_API_KEY, model="text-embedding
|
|
22 |
# Create Embeddings for Searching the Splits
|
23 |
persist_directory = './chroma/'
|
24 |
|
25 |
-
def initialize():
|
26 |
-
|
27 |
-
|
28 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
|
|
|
|
|
30 |
splits = gen_splits.gen_splits()
|
31 |
client = chromadb.Client()
|
32 |
collection = client.create_collection(name="docs")
|
33 |
-
|
|
|
34 |
# store each document in a vector embedding database
|
35 |
for i, d in enumerate(splits):
|
36 |
-
|
37 |
-
|
38 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
|
40 |
return collection
|
41 |
-
|
42 |
|
43 |
if __name__ == "__main__":
|
44 |
vectordb = initialize()
|
|
|
22 |
# Create Embeddings for Searching the Splits
|
23 |
persist_directory = './chroma/'
|
24 |
|
25 |
+
# def initialize():
|
26 |
+
# # splits = gen_splits.gen_splits()
|
27 |
+
# # vectordb = Chroma.from_documents(documents=splits, persist_directory=persist_directory, embedding=embedding_model)
|
28 |
+
# # vectordb.persist()
|
29 |
+
|
30 |
+
# splits = gen_splits.gen_splits()
|
31 |
+
# client = chromadb.Client()
|
32 |
+
# collection = client.create_collection(name="docs")
|
33 |
+
# print(splits)
|
34 |
+
# # store each document in a vector embedding database
|
35 |
+
# for i, d in enumerate(splits):
|
36 |
+
# response = ollama.embeddings(model="mxbai-embed-large", prompt=d.page_content)
|
37 |
+
# embedding = response["embedding"]
|
38 |
+
# collection.add(ids=[str(i)],embeddings=[embedding], documents=[d])
|
39 |
+
|
40 |
+
# return collection
|
41 |
|
42 |
+
|
43 |
+
def initialize():
|
44 |
splits = gen_splits.gen_splits()
|
45 |
client = chromadb.Client()
|
46 |
collection = client.create_collection(name="docs")
|
47 |
+
|
48 |
+
|
49 |
# store each document in a vector embedding database
|
50 |
for i, d in enumerate(splits):
|
51 |
+
success = False
|
52 |
+
attempts = 0
|
53 |
+
max_attempts = 5
|
54 |
+
|
55 |
+
while not success and attempts < max_attempts:
|
56 |
+
try:
|
57 |
+
response = ollama.embeddings(model="mxbai-embed-large", prompt=d.page_content)
|
58 |
+
embedding = response["embedding"]
|
59 |
+
collection.add(ids=[str(i)], embeddings=[embedding], documents=[d])
|
60 |
+
success = True
|
61 |
+
except httpx.ConnectError as e:
|
62 |
+
attempts += 1
|
63 |
+
print(f"Connection failed (attempt {attempts}): {e}")
|
64 |
+
time.sleep(2) # retry after waiting for 2 seconds
|
65 |
|
66 |
return collection
|
67 |
+
|
68 |
|
69 |
if __name__ == "__main__":
|
70 |
vectordb = initialize()
|