update v3 single gpu rule yaml;

ktransformers/local_chat.py

@@ -32,8 +32,7 @@ custom_models = {
 ktransformer_rules_dir = os.path.dirname(os.path.abspath(__file__)) + "/optimize/optimize_rules/"
 default_optimize_rules = {
     "DeepseekV2ForCausalLM": ktransformer_rules_dir + "DeepSeek-V2-Chat.yaml",
-    # "DeepseekV3ForCausalLM": ktransformer_rules_dir + "DeepSeek-V2-Chat.yaml",
-    "DeepseekV3ForCausalLM": ktransformer_rules_dir + "DeepSeek-V3-Chat-multi-gpu.yaml",
+    "DeepseekV3ForCausalLM": ktransformer_rules_dir + "DeepSeek-V3-Chat.yaml",
     "Qwen2MoeForCausalLM": ktransformer_rules_dir + "Qwen2-57B-A14B-Instruct.yaml",
     "LlamaForCausalLM": ktransformer_rules_dir + "Internlm2_5-7b-Chat-1m.yaml",
     "MixtralForCausalLM": ktransformer_rules_dir + "Mixtral.yaml",

ktransformers/optimize/optimize_rules/DeepSeek-V3-Chat.yaml

 - match:
-    class: ktransformers.models.modeling_deepseek.DeepseekV3YarnRotaryEmbedding
+    class: ktransformers.models.modeling_deepseek_v3.DeepseekV3RotaryEmbedding
   replace:
-    class: ktransformers.operators.RoPE.YarnRotaryEmbedding
+    class: ktransformers.operators.RoPE.RotaryEmbeddingV3
     kwargs:
       generate_device: "cuda"
       prefill_device: "cuda"
@@ -17,12 +17,19 @@
       prefill_op: "KLinearTorch"
 - match:
     name: "^model\\.layers\\..*\\.mlp$"
-    class: ktransformers.models.modeling_deepseek.DeepseekV2MoE
+    class: ktransformers.models.modeling_deepseek_v3.DeepseekV3MoE
   replace:
-    class: ktransformers.operators.experts.KDeepseekV2MoE     # mlp module with custom forward function
+    class: ktransformers.operators.experts.KDeepseekV3MoE     # mlp module with custom forward function
     kwargs:
       generate_device: "cuda"
       prefill_device: "cuda"
+- match:
+    class: ktransformers.models.modeling_deepseek_v3.MoEGate
+  replace:
+    class: ktransformers.operators.gate.KMoEGate
+    kwargs:
+      generate_device: "cuda:0"
+      prefill_device: "cuda:0"
 - match:
     name: "^model\\.layers\\..*\\.mlp\\.experts$"
   replace: