Finetuning hyperparameters and final training loss

#1
by Viewegger - opened

Hello,

I would like to make similar finetune on 3090 RTX card, may I ask you what hyperparameters did you use and what final training loss you were able to achieve?

Also is the model good considering only about 20k examples? - I don't speak any Urdu so I cannot judge :)

Sign up or log in to comment