diff --git a/vllm/model_executor/models/deepseek.py b/vllm/model_executor/models/deepseek.py
index c6421143dd68..bcde432280a8 100644
--- a/vllm/model_executor/models/deepseek.py
+++ b/vllm/model_executor/models/deepseek.py
@@ -45,7 +45,8 @@
 from vllm.model_executor.layers.rotary_embedding import get_rope
 from vllm.model_executor.layers.vocab_parallel_embedding import (
     ParallelLMHead, VocabParallelEmbedding)
-from vllm.model_executor.model_loader.weight_utils import default_weight_loader
+from vllm.model_executor.model_loader.weight_utils import (
+    default_weight_loader, maybe_remap_kv_scale_name)
 from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.sequence import IntermediateTensors
 
@@ -414,6 +415,11 @@ def load_weights(self, weights: Iterable[Tuple[str,
                     continue
                 if is_pp_missing_parameter(name, self):
                     continue
+                if name.endswith("scale"):
+                    # Remapping the name of FP8 kv-scale.
+                    name = maybe_remap_kv_scale_name(name, params_dict)
+                    if name is None:
+                        continue
                 param = params_dict[name]
                 weight_loader = param.weight_loader
                 weight_loader(param, loaded_weight, shard_id)
@@ -428,6 +434,10 @@ def load_weights(self, weights: Iterable[Tuple[str,
                     continue
                 if is_pp_missing_parameter(name, self):
                     continue
+                # Remapping the name of FP8 kv-scale.
+                name = maybe_remap_kv_scale_name(name, params_dict)
+                if name is None:
+                    continue
                 param = params_dict[name]
                 weight_loader = getattr(param, "weight_loader",
                                         default_weight_loader)