fine-tuning an LLM

Published 2025-07-16 • Updated 2025-10-11

I’ve been playing around with fine-tuning a bit, both through OpenAI and locally. I may describe the process at some point. It’s both rewarding and frustrating, unsurprisingly.

To fine-tune locally, I had to rent a GPU through Celium, a subnet on the Bittensor network, since my laptop GPU is nowhere near powerful enough. Bittensor is an interesting concept; I’ve explored it a bit and may write about it someday.