google
/

gemma-3-12b-pt

model.config.vocab_size has a wrong value compared to the actual vocab size in the tokenizer

#2 opened about 2 months ago by