[FP8 KV Cache] Avoid KeyError at loading pre-quantized FP8 model with kv_scale (#1559)

python/sglang/srt/models/llama.py

@@ -400,6 +400,9 @@ class LlamaForCausalLM(nn.Module):
                 # Skip loading extra bias for GPTQ models.
                 if name.endswith(".bias") and name not in params_dict:
                     continue
+                # Skip loading kv_scale from ckpts towards new design.
+                if name.endswith(".kv_scale") and name not in params_dict:
+                    continue
                 param = params_dict[name]
                 weight_loader = getattr(param, "weight_loader", default_weight_loader)
                 weight_loader(param, loaded_weight)