{ "cells": [ { "cell_type": "markdown", "id": "1c6056e6-86ea-4c92-9e95-a64b38d6f4c9", "metadata": {}, "source": [ "# Build a self-supervised learning pipeline\n", "\n", "This notebook will explain how to use the selfeeg library to build a self-supervised learning pipeline. \n", "\n", "To summarize, typical steps include:\n", "\n", "1. [PRETRAINING](#Pretraining-Phase):\n", " 1. [define the pretraining dataloaders](#Define-Dataloaders)\n", " 2. [define the augmenter](#Define-the-data-augmenter)\n", " 3. [define the pretraining model](#Define-pretraining-model-and-other-training-objects) (and optional training elements)\n", " 4. [pretrain the model](#Pretrain-the-model)\n", "2. [FINE-TUNING](#Fine-tuning-Phase)\n", " 1. [define the fine-tuning dataloaders](#Define-fine-tuning-dataloaders)\n", " 2. [define the fine-tuning model](#Define-fine-tuning-model-and-other-training-objects) (and optional training elements)\n", " 3. [transfer the pretrained encoder's weights](#Define-fine-tuning-model-and-other-training-objects)\n", " 4. [fine-tune the model](#Fine-tuning)\n", "3. [FINAL EVALUATION](#Evaluate-fine-tuned-model)\n", "\n", "To better understand how the **dataloading** and **augmentations** module work, check the respective introductory notebooks" ] }, { "cell_type": "markdown", "id": "2813d14e-e23b-4cec-8490-6e92ed1bc02a", "metadata": {}, "source": [ "First, let's import all the packages necessary to run this notebook.\n", "\n", "