Tokenizer `model_max_length` does not match DeepSeek-Prover-V1.5-Base

#3
by 0xekez - opened

DeepSeek-Prover-V1.5-Base (this model's base model) sets the model_max_length field to 16384 in tokenizer_config.json, in this repo it is 4096 . Should this repo's value be the same as the base model?

Sign up or log in to comment