Chapter 15.1 - Chatting
The content will be uploaded soon
Chapter-15.2---Chatting Dataset and loader
The content will be uploaded soon
Chapter 15.3 - Chat Dataset
When dealing with a multi-turn chatting dataset like UltraChat, we have to prepare the data in a very specific way before our GPT-2 model can learn from it.
Chapter 15.4 - Collate function
The Collate Function (r_collate.py in our architecture) acts as the final preparation stage before a batch of conversations is sent into the neural network for training. It performs two critical jobs: Padding and Loss Masking.