I assumed it's qwen3 but it seems to be something other. Because there is no such thing as qwen3 in llama.cpp tokenizers.
PreTrainedTokenizerFast
Not worked. I found myself it's qwen2.
qwen2
· Sign up or log in to comment
