How-To Instruct Fine-Tuning Falcon-7B [Google Colab Included]

This Tutorial shows how to fine-tune the recent Falcon-7b model on a single Google colab and turn it into a chatbot

We will leverage PEFT library from Hugging Face ecosystem, as well as QLoRA for more memory efficient finetuning.

We will use the Guanaco dataset, which is a clean subset of the OpenAssistant dataset adapted to train general purpose chatbots.

Here we will use the SFTTrainer from TRL library that gives a wrapper around transformers Trainer to easily fine-tune models …

chatbot colab dataset ecosystem face falcon fine-tuning finetuning google how-to hugging face library memory openassistant shows tutorial