You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 

58 lines
1.7 KiB

WEBVTT
00:01.160 --> 00:05.570
And would you believe at this point you're 55% of the way along the journey?
00:05.780 --> 00:09.890
Uh, it's been a while since I've thrown that stat out there, but congratulations.
00:09.890 --> 00:11.120
That's really awesome.
00:11.210 --> 00:14.270
Uh, so we have made progress with rag.
00:14.270 --> 00:18.560
We're taking it slowly because it's about to get real as we get into vector databases.
00:18.560 --> 00:24.920
But at this point, uh, not only do you have the sort of foundational understanding of why we're going
00:24.920 --> 00:31.310
to be using vectors, but you can now use Lang chain to load in documents, to split them, to add in
00:31.310 --> 00:32.120
metadata.
00:32.120 --> 00:36.620
And hopefully you've played around with that and you've satisfied yourself about how it's working and
00:36.620 --> 00:39.020
you're ready for the real deal.
00:39.020 --> 00:46.070
We next time are going to take these chunks and convert them into vectors using OpenAI's OpenAI embeddings,
00:46.070 --> 00:49.280
which was one of the encoding llms that we talked about.
00:49.310 --> 00:55.640
We're then going to store the vectors in an open source vector data store called chroma, which is an
00:55.640 --> 00:59.090
extremely popular open source vector database.
00:59.090 --> 01:00.080
And it's terrific.
01:00.080 --> 01:04.790
And we're going to enjoy putting our vectors in there, because we're then going to visualize them and
01:04.790 --> 01:10.010
see them and get a sense of what does it even mean to have a vector in a database.
01:10.010 --> 01:11.930
So lots to be done.
01:11.930 --> 01:13.790
I will see you next time.