From the uDemy course on LLM engineering.
https://www.udemy.com/course/llm-engineering-master-ai-and-large-language-models
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.
82 lines
2.2 KiB
82 lines
2.2 KiB
WEBVTT |
|
|
|
00:00.920 --> 00:04.850 |
|
So we've covered steps 1 to 4 of the five step strategy. |
|
|
|
00:04.850 --> 00:10.010 |
|
And that brings us to step five, which is productionize the one that maybe you haven't thought as much |
|
|
|
00:10.010 --> 00:11.240 |
|
about to date. |
|
|
|
00:11.240 --> 00:13.520 |
|
And this is about saying we built our model. |
|
|
|
00:13.520 --> 00:16.250 |
|
It's time for prime time. |
|
|
|
00:16.730 --> 00:23.810 |
|
We typically will start by defining an API that can be used to call our model to call our code. |
|
|
|
00:23.810 --> 00:32.570 |
|
That itself is either using a open source model or making a call to a frontier model with with various |
|
|
|
00:32.780 --> 00:36.680 |
|
scaffolding around it, like rag or prompt engineering. |
|
|
|
00:37.190 --> 00:42.500 |
|
We then decide how are we going to host and deploy our model, our API. |
|
|
|
00:42.530 --> 00:48.410 |
|
We address things like how it's going to be monitored, the information security concerns, how will |
|
|
|
00:48.410 --> 00:49.580 |
|
it scale up? |
|
|
|
00:49.730 --> 00:56.750 |
|
We then measure the business metrics that we had identified all the way back in step one and continuously |
|
|
|
00:56.750 --> 01:02.390 |
|
retrain, measure performance, and improve our model on an ongoing basis. |
|
|
|
01:02.420 --> 01:06.320 |
|
These are all part of the Productionize activity. |
|
|
|
01:07.820 --> 01:14.660 |
|
And so with that, where are we in our business problem that we've taken on to predict product prices. |
|
|
|
01:14.660 --> 01:17.600 |
|
So the first step understand we've done it. |
|
|
|
01:17.630 --> 01:18.410 |
|
We understood it. |
|
|
|
01:18.440 --> 01:19.850 |
|
We've talked about metrics. |
|
|
|
01:19.850 --> 01:21.170 |
|
We know what we've got to do. |
|
|
|
01:21.200 --> 01:23.900 |
|
Preparation is what we're doing right now. |
|
|
|
01:23.900 --> 01:26.630 |
|
We are in the middle of preparing our data set. |
|
|
|
01:26.630 --> 01:30.290 |
|
And then we need to select our model or models. |
|
|
|
01:30.290 --> 01:32.300 |
|
We need to customize. |
|
|
|
01:32.390 --> 01:35.060 |
|
And then finally we need to productionize. |
|
|
|
01:35.060 --> 01:38.060 |
|
So we will be going through all of those steps. |
|
|
|
01:38.150 --> 01:43.400 |
|
But we will be first of all finishing the second step preparation. |
|
|
|
01:43.400 --> 01:49.520 |
|
And with that let's go right now back to JupyterLab and get to preparing.
|
|
|