Question about tokenizer.json size

#8
by eatbeans2 - opened

There was a discussion in this reddit post about many finetunes having an incorrect and oversized tokenizer.json files. I noticed that the tokenizer for this model is almost 2MB larger than the 2407 version. Is this expected?

Sign up or log in to comment