DOP 222: Finding Performance Bottlenecks With Ddosify
DevOps Paradox
00:00
How to Pre-Train a Model
The first person we had on that actually understands this from an academic level. We will use the most of the layers of GPT models and pre-train only the last layers. So the model can understand the metrics from all our customers' data, not only from the Wikipedia site. The OpenAI current model is very large white, we want to get into the narrow state.
Transcript
Play full episode
