Why this model uses a different tokenizer from other variants (0.5B/1.8B/4B)?

#6
by J22 - opened

Are you going to release a new 7B model?

Sign up or log in to comment