Update rag.py
Browse files
rag.py
CHANGED
@@ -17,9 +17,10 @@ from pymongo import MongoClient
|
|
17 |
RAG_CHROMA = "Chroma"
|
18 |
RAG_MONGODB = "MongoDB"
|
19 |
|
20 |
-
PDF_URL
|
21 |
-
WEB_URL
|
22 |
-
|
|
|
23 |
|
24 |
YOUTUBE_DIR = "/data/youtube"
|
25 |
CHROMA_DIR = "/data/chroma"
|
@@ -47,7 +48,7 @@ def document_loading():
|
|
47 |
docs.extend(loader.load())
|
48 |
|
49 |
# YouTube
|
50 |
-
loader = GenericLoader(YoutubeAudioLoader([
|
51 |
OpenAIWhisperParser())
|
52 |
docs.extend(loader.load())
|
53 |
|
|
|
17 |
RAG_CHROMA = "Chroma"
|
18 |
RAG_MONGODB = "MongoDB"
|
19 |
|
20 |
+
PDF_URL = "https://arxiv.org/pdf/2303.08774.pdf"
|
21 |
+
WEB_URL = "https://openai.com/research/gpt-4"
|
22 |
+
YOUTUBE_URL_1 = "https://www.youtube.com/watch?v=--khbXchTeE"
|
23 |
+
YOUTUBE_URL_2 = "https://www.youtube.com/watch?v=hdhZwyf24mE"
|
24 |
|
25 |
YOUTUBE_DIR = "/data/youtube"
|
26 |
CHROMA_DIR = "/data/chroma"
|
|
|
48 |
docs.extend(loader.load())
|
49 |
|
50 |
# YouTube
|
51 |
+
loader = GenericLoader(YoutubeAudioLoader([YOUTUBE_URL_1, YOUTUBE_URL_2], YOUTUBE_DIR),
|
52 |
OpenAIWhisperParser())
|
53 |
docs.extend(loader.load())
|
54 |
|