Hi, just wanted to make a request — could you create a post explaining how we moved from full fine-tuning to using adapters, and then to PEFT for fine-tuning? I think this progression is crucial for anyone who really wants to understand fine-tuning in depth. And not to forget — this Part 2 is just as good as Part 1
Hi, just wanted to make a request — could you create a post explaining how we moved from full fine-tuning to using adapters, and then to PEFT for fine-tuning? I think this progression is crucial for anyone who really wants to understand fine-tuning in depth. And not to forget — this Part 2 is just as good as Part 1
Kudos on the informative article Shivani ! How do you manage to overcome the real-world issues such as
1. model metrics not meeting the defined specifications
2. messy inputs in production for eg PDF's with unclear figures as input for a doc processing or ocr usecase