[NVIDIA] Auto detect modelopt quant and fix DSR1-FP4 weight loading (#22073)

vllm/config.py

@@ -1108,6 +1108,21 @@ class ModelConfig:
         if quant_cfg is None:
             # compressed-tensors uses a "compression_config" key
             quant_cfg = getattr(self.hf_config, "compression_config", None)
+
+        else:
+            # Set quant_method for ModelOpt models.
+            producer_name = quant_cfg.get("producer", {}).get("name")
+            if producer_name == "modelopt":
+                quant_algo = quant_cfg.get("quantization",
+                                           {}).get("quant_algo")
+                if quant_algo == "FP8":
+                    quant_cfg["quant_method"] = "modelopt"
+                elif quant_algo == "NVFP4":
+                    quant_cfg["quant_method"] = "modelopt_fp4"
+                elif quant_algo is not None:
+                    raise ValueError(
+                        f"Unknown ModelOpt quant algo: {quant_algo}")
+
         return quant_cfg
 
     def _verify_quantization(self) -> None:

vllm/model_executor/layers/fused_moe/layer.py

@@ -919,9 +919,13 @@ class FusedMoE(torch.nn.Module):
         elif shard_id == "w2":
             param_data[expert_id] = loaded_weight
 
-    def _load_w13_weight_scale(self, shard_dim: int,
-                               loaded_weight: torch.Tensor,
-                               param: torch.Tensor, tp_rank: int):
+    def _load_combined_w13_weight_scale(self, shard_dim: int,
+                                        loaded_weight: torch.Tensor,
+                                        param: torch.Tensor, tp_rank: int):
+        """
+        Load w13 weight scales assuming that w1 weight scales and w3 weight
+        scales are stored in the same loaded_weight tensor.
+        """
         shard_size = param.shape[shard_dim]
         loaded_weight = loaded_weight.narrow(shard_dim, shard_size * tp_rank,
                                              shard_size)
@@ -1168,24 +1172,43 @@ class FusedMoE(torch.nn.Module):
             uses_weight_scale_2 = self.quant_method.uses_weight_scale_2_pattern(
             )
 
-            # For per-tensor, FP4 uses "weight_scale_2", FP8 uses "weight_scale"
-            per_tensor_conditions = (
-                "weight_scale_2" in weight_name if uses_weight_scale_2 else
-                "weight_scale" in weight_name) or "input_scale" in weight_name
-
-            if "w13_weight_scale" in weight_name:
-                self._load_w13_weight_scale(shard_dim=shard_dim,
-                                            loaded_weight=loaded_weight,
-                                            param=param,
-                                            tp_rank=self.tp_rank)
-            elif per_tensor_conditions:
+            # Call _load_per_tensor_weight_scale() to load per-tensor (scalar)
+            # weights scales.
+            # Input scales are always per-tensor.
+            # Weight scales: FP4 uses "weight_scale_2" and FP8 uses
+            # "weight_scale" for per-tensor scales.
+            is_per_tensor = ("weight_scale_2" in weight_name
+                             if uses_weight_scale_2 else "weight_scale"
+                             in weight_name) or "input_scale" in weight_name
+            if is_per_tensor:
                 self._load_per_tensor_weight_scale(
                     shard_id=shard_id,
                     param=param,
                     loaded_weight=loaded_weight,
                     expert_id=expert_id,
                 )
-            elif "weight" in weight_name:
+                return True if return_success else None
+
+            # If the weight is w13_weight_scale and w13_weight_scales are
+            # combined into single loaded_weight, call
+            # _load_combined_w13_weight_scale() to load it.
+            # This is checked by comparing the hidden_out dims of the
+            # loaded_weight and the param.
+            if "w13_weight_scale" in weight_name:
+                loaded_weight_hidden_out = loaded_weight.shape[-2]
+                param_hidden_out = param.data.shape[-2] * self.tp_size
+                if loaded_weight_hidden_out == param_hidden_out:
+                    self._load_combined_w13_weight_scale(
+                        shard_dim=shard_dim,
+                        loaded_weight=loaded_weight,
+                        param=param,
+                        tp_rank=self.tp_rank,
+                    )
+                    return True if return_success else None
+
+            # For other weights, call _load_model_weight_or_group_weight_scale()
+            # to load it.
+            if "weight" in weight_name:
                 self._load_model_weight_or_group_weight_scale(
                     shard_id=shard_id,
                     shard_dim=shard_dim,

vllm/transformers_utils/config.py

@@ -449,6 +449,20 @@ def get_config(
         model_type = MODEL_FOR_CAUSAL_LM_MAPPING_NAMES[config.model_type]
         config.update({"architectures": [model_type]})
 
+    # ModelOpt 0.31.0 and after saves the quantization config in the model
+    # config file.
+    quantization_config = config_dict.get("quantization_config", None)
+
+    # ModelOpt 0.29.0 and before saves the quantization config in a separate
+    # "hf_quant_config.json" in the same directory as the model config file.
+    if quantization_config is None \
+        and file_or_path_exists(model, "hf_quant_config.json", revision):
+        quantization_config = get_hf_file_to_dict("hf_quant_config.json",
+                                                  model, revision)
+
+    if quantization_config is not None:
+        config.quantization_config = quantization_config
+
     if hf_overrides_kw:
         logger.debug("Overriding HF config with %s", hf_overrides_kw)
         config.update(hf_overrides_kw)