Hello everyone!

So before, the base model finetuned for KobiMusic ran off what i thought was gpt-2 medium, but it was actually the gpt-2 small model, just 125 million parameters. For its intended purpose, text generation, go to the huggingface interference and yeah, thats not the same gpt that can do your homework.

Anyways, KobiMusic will be getting gpt-2 medium which is 2x the size of gpt-2-small, in probably a week or less potentially