Experiments in Open Source LLMs for Augmented and Alternative Communication

These are the files for the PyData Vancouver talk.

The order on which the experiments goes is:

1. Sample, see aac-speech-standalone.
2. Annotate, see data.
3. LoRA fine-tune and export, see src.
4. Quantize for C++ execution, see RedPajamaCpp.md.