Comment by anon373839

If you want to experiment with fine tuning small models, you can actually do that with free Google Colab instances. Unsloth.ai has written a bunch of guides on this.