text-generation-inference

fix: tokenizer config should use local model path when possible

#1518

Merged