Text Generation
Transformers
PyTorch
English
gpt2
causal-lm
text-generation-inference

New checkpoint - what's the difference?

#1
by eugeneware - opened

Thanks so much for sharing this work with the community. I notice that a new checkpoint was uploaded. Do you know what the difference is? If we have a trained LoRA adapter on the previous model, would we expect this to still work? Or is the model significantly different?

@eugeneware sure! this is an updated checkpoint with no architecture difference. You probably want to update checkpoint and re-train LoRA.

Thanks @daria-soboleva - will do. Keep up the great work πŸ‘πŸ»

eugeneware changed discussion status to closed

Sign up or log in to comment