Spaces:
Running
Running
Get API Key and host URL from env vars
Browse files- .gitignore +1 -1
- create_index.py +36 -8
.gitignore
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
__pycache__
|
2 |
*.npy
|
3 |
-
*.
|
|
|
1 |
__pycache__
|
2 |
*.npy
|
3 |
+
*.json
|
create_index.py
CHANGED
@@ -1,18 +1,46 @@
|
|
|
|
|
|
1 |
from encoder import encode
|
2 |
|
3 |
import numpy as np
|
4 |
import tqdm
|
5 |
from datasets import load_dataset
|
6 |
|
7 |
-
batch_size = 64
|
8 |
|
9 |
-
|
|
|
|
|
10 |
|
11 |
-
|
|
|
|
|
12 |
|
13 |
-
|
14 |
-
|
15 |
-
batch_embeddings = encode(batch_answers)
|
16 |
-
answer_embeddings[i:i+batch_size] = batch_embeddings
|
17 |
|
18 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import argparse
|
2 |
+
import os
|
3 |
from encoder import encode
|
4 |
|
5 |
import numpy as np
|
6 |
import tqdm
|
7 |
from datasets import load_dataset
|
8 |
|
|
|
9 |
|
10 |
+
def get_embeddings_and_payload(dataset, batch_size):
|
11 |
+
answer_embeddings = np.zeros(shape=(len(dataset), 384), dtype=np.float32)
|
12 |
+
answer_texts = []
|
13 |
|
14 |
+
for i in tqdm.tqdm(range(0, len(dataset), batch_size), desc='Extracting vectors'):
|
15 |
+
batch_answers = dataset['answer'][i:i+batch_size]
|
16 |
+
answer_texts.extend([{"text": answer in batch_answers}])
|
17 |
|
18 |
+
batch_embeddings = encode(batch_answers)
|
19 |
+
answer_embeddings[i:i+batch_size] = batch_embeddings
|
|
|
|
|
20 |
|
21 |
+
return answer_embeddings, answer_texts
|
22 |
+
|
23 |
+
|
24 |
+
def main(args):
|
25 |
+
if args.host_url is None:
|
26 |
+
print("You must specify a host URL either by passing it as a command line argument with `--host-url` or as an environment variable named `QDRANT_HOST_URL`")
|
27 |
+
exit()
|
28 |
+
if args.api_key is None:
|
29 |
+
print("You must specify an API key either by passing it as a command line argument with `--api-key` or as an environment variable named `QDRANT_API_KEY`")
|
30 |
+
exit()
|
31 |
+
|
32 |
+
dataset = load_dataset('wiki_qa', split='train+test+validation')
|
33 |
+
embeddings, payload_data = get_embeddings_and_payload(
|
34 |
+
dataset, args.batch_size)
|
35 |
+
|
36 |
+
|
37 |
+
if __name__ == "__main__":
|
38 |
+
ap = argparse.ArgumentParser(prog="Create index on on Qdrant Cloud")
|
39 |
+
ap.add_argument("-b", "--batch-size", type=int, default=64,
|
40 |
+
help="Batch size when extracting vectors")
|
41 |
+
ap.add_argument("-u", "--host-url", type=str,
|
42 |
+
default=os.environ['QDRANT_HOST_URL', None], help="Host URL from Qdrant Cloud")
|
43 |
+
ap.add_argument("-k", "--api-key", type=str,
|
44 |
+
default=os.environ['QDRANT_API_KEY', None], help="API key from Qdrant Cloud")
|
45 |
+
args = ap.parse_args()
|
46 |
+
main(args)
|