HuggingFace transformers processor validates tokenizer type by checking
the class name. When vLLM creates a CachedTokenizer with a modified
class name (e.g., 'CachedQwen2TokenizerFast'), the processor type check
fails with TypeError.
This fix preserves the original tokenizer class name and qualname in
CachedTokenizer, ensuring compatibility with HuggingFace transformers
processor type checking.
Fixes#31080
Signed-off-by: Claude <noreply@anthropic.com>
Signed-off-by: majiayu000 <1835304752@qq.com>
Signed-off-by: Julien Denize <julien.denize@mistral.ai>
Signed-off-by: Julien Denize <40604584+juliendenize@users.noreply.github.com>
Signed-off-by: Mickael Seznec <mickael@mistral.ai>
Signed-off-by: Roger Wang <hey@rogerw.io>
Co-authored-by: Roger Wang <hey@rogerw.io>
Co-authored-by: Mickael Seznec <mickael@mistral.ai>
