File size: 432 Bytes
57bdca5 |
1 2 3 4 5 6 7 |
We'll see how to build upon transformers and extend the framework with your hooks and custom code. We will illustrate all of this on a ResNet model, by wrapping the ResNet class of the timm library into a [PreTrainedModel]. Writing a custom configuration Before we dive into the model, let's first write its configuration. The configuration of a model is an object that will contain all the necessary information to build the model. |