harshbhatia66 / BallsDeepLearning

0 stars 1 forks source link

Finetuning pygmalion #10

Closed harshbhatia66 closed 8 months ago

harshbhatia66 commented 1 year ago

https://github.com/PygmalionAI/training-code, there is an existing training git code which uses lora. The dataset orientation needs to be formatted into alpacca style.
https://github.com/oobabooga/text-generation-webui/blob/main/docs/Training-LoRAs.md#format-files