Fix/nemotron mtp quantaized (#19433)

2026-03-03 01:07:46 -08:00
parent af0d35b224
commit 4c95953b77
5 changed files with 73 additions and 3 deletions
@@ -665,7 +665,10 @@ class ModelConfig:
            quant_cfg = quant_cfg.to_dict()
        if quant_cfg is not None:
            # Identify modelopt quantization
-            if "quant_method" not in quant_cfg:
+            if (
+                "quant_method" not in quant_cfg
+                or quant_cfg["quant_method"] == "modelopt"
+            ):
                parsed_cfg = self._parse_modelopt_quant_config(
                    {"quantization": quant_cfg}
                )