fix: change tokenizer_class from TokenizersBackend to Qwen2Tokenizer for vLLM compat

tokenizer_config.json

@@ -23,9 +23,9 @@
   "pad_token": "<|endoftext|>",
   "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
   "split_special_tokens": false,
-  "tokenizer_class": "TokenizersBackend",
+  "tokenizer_class": "Qwen2Tokenizer",
   "unk_token": null,
   "video_token": "<|video_pad|>",
   "vision_bos_token": "<|vision_start|>",
   "vision_eos_token": "<|vision_end|>"
-}
+}