Do not call tokenizer.save_pretrained(), to avoid overwriting tokenizer_config.json. (#2410)

`model.save()` has already correctly saved the tokenizer.

Signed-off-by: ZX-ModelCloud <zx@modelcloud.ai>

Author: ZX-ModelCloud

tests/models/model_test.py

@@ -949,7 +949,6 @@ def quantModel(self, model_id_or_path, trust_remote_code=False, dtype="auto", ne
                     self.clear_directory(path)
 
                     model.save(path)
-                    tokenizer.save_pretrained(path)
                     self._print_post_quant_artifacts(path)
 
                     reuse_candidates = {}