This is a project for fine tuning an LLM on ThatsGroes/synthetic-dialog-summaries-processed-clean-chatml
The main fine tuning script is found in train.py. This script takes a dictionary
of configurations as input. See examples in the training_configs directory.
The scripts starting with train_ are scripts that load a specific config file
and calls the train() function in train.py.
See my LLM fine tuning notebook for a walkthrough of the training code and the related concepts: LLM fine tuning notebook