skip is_v32

vllm/model_executor/layers/fused_moe/fused_moe.py

@@ -7,7 +7,7 @@ import os
 # torch.compile needs typing.List. It will fail torch.library.infer_schema
 # otherwise
 from typing import List  # noqa: UP035
-from typing import Any, Callable, Optional, Union, List
+from typing import Any, Callable, Optional, Union, List, Dict
 
 
 import torch

vllm/model_executor/models/deepseek_v2.py

@@ -938,7 +938,8 @@ class DeepseekV2MLAAttention(nn.Module):
             mscale = yarn_get_mscale(scaling_factor, float(mscale_all_dim))
             self.scaling = self.scaling * mscale * mscale
 
-        self.is_v32 = hasattr(config, "index_topk")
+        # self.is_v32 = hasattr(config, "index_topk")
+        self.is_v32 = False
 
         if self.is_v32:
             self.indexer = Indexer(vllm_config, config, hidden_size,
@@ -1116,7 +1117,8 @@ class DeepseekV2Model(nn.Module):
         self.config = config
 
         self.vocab_size = config.vocab_size
-        self.is_v32 = hasattr(config, "index_topk")
+        # self.is_v32 = hasattr(config, "index_topk")
+        self.is_v32 = False
         if self.is_v32:
             topk_tokens = config.index_topk
             topk_indices_buffer = torch.empty(