Merge pull request #911 from kvcache-ai/patch_v0.2.3post2

🔧 update multi-gpu-fp8-linear and multi-gpu marlin yaml

Author: KMSorSMS

ktransformers/optimize/optimize_rules/DeepSeek-V3-Chat-multi-gpu-fp8-linear-ggml-experts.yaml

@@ -10,15 +10,15 @@
     name: "^model\\.layers\\.(0|[1-9]|[12][0-9])\\."
     class: ktransformers.models.modeling_deepseek_v3.DeepseekV3RotaryEmbedding
   replace:
-    class: ktransformers.operators.RoPE.KMoEGateDeepSeekV3
+    class: ktransformers.operators.RoPE.YarnRotaryEmbeddingV3
     kwargs:
       generate_device: "cuda:0"
       prefill_device: "cuda:0"
 - match:
     name: "^model\\.layers\\.([3456][0-9])\\."
     class: ktransformers.models.modeling_deepseek_v3.DeepseekV3RotaryEmbedding
   replace:
-    class: ktransformers.operators.RoPE.KMoEGateDeepSeekV3
+    class: ktransformers.operators.RoPE.YarnRotaryEmbeddingV3
     kwargs:
       generate_device: "cuda:1"
       prefill_device: "cuda:1"

ktransformers/optimize/optimize_rules/DeepSeek-V3-Chat-multi-gpu-marlin.yaml

@@ -10,7 +10,7 @@
     name: "^model\\.layers\\.(0|[1-9]|[12][0-9])\\."
     class: ktransformers.models.modeling_deepseek_v3.DeepseekV3RotaryEmbedding
   replace:
-    class: ktransformers.operators.RoPE.KMoEGateDeepSeekV3
+    class: ktransformers.operators.RoPE.YarnRotaryEmbeddingV3
     kwargs:
       generate_device: "cuda:0"
       prefill_device: "cuda:0"