request for training notebook to finetune falcon 7b

#1
by g30rv17ys - opened

@ehartford .

hope you could share the .ipynb for fine-tuning the falcon-7b model.
it would be great if you could upload it as a colab notebook.
also have u explored QLora?

Cognitive Computations org
β€’
edited May 30, 2023

No I used axolotl.
https://github.com/OpenAccess-AI-Collective/axolotl

I don't need qLoRA for this small training job. When I train 65b then I'll use qLoRA.

Absolutely love this idea! Any chance you could release the training set, @ehartford ? :)

Cognitive Computations org

I'm working on it.

can you please describe how you managed to fine tune it ?

can you please describe how you managed to fine tune it ?

Do you mean his parameters to axolotl?

Cognitive Computations org

Did you use:
python scripts/finetune.py examples/falcon/config-7b.yml

or:
accelerate launch scripts/finetune.py examples/falcon/config-7b.yml

Sign up or log in to comment