Text Generation
Transformers
PyTorch
English
llama
Inference Endpoints
text-generation-inference

Training tokens

#34
by adminscholaro - opened

Hello! I was wondering how many tokens StableBeluga2 is trained on. I've read that it is trained on 10% of the Ocra dataset, which is 32 T tokens as far as I've understood, it is correct to assume StableBeluga2 is trained on 3.2 T tokens then?

Sign up or log in to comment