This repo contains various experiments done in April-May 2024 tuning open and private LLMs to write in the style of a particular author. A sample of about 3000 diary entries written at different ages was used as training data.
The objective, beyond the experiment in itself, was to undersand how fine-tuning LLMs work in different environments: AWS SageMaker Studio using GPU instances, SageMarker only for the training runs, AWS Bedrock, locally in an Apple M1, OpenAI, Azure.