Remove unused kwargs from model definitions (#13555)

vllm/model_executor/models/minicpm3.py

@@ -29,7 +29,7 @@ import torch
 from torch import nn
 from transformers import PretrainedConfig
 
-from vllm.attention import Attention, AttentionMetadata
+from vllm.attention import Attention
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_tensor_model_parallel_world_size
 from vllm.model_executor.layers.layernorm import RMSNorm
@@ -129,8 +129,6 @@ class MiniCPM3Attention(nn.Module):
         self,
         positions: torch.Tensor,
         hidden_states: torch.Tensor,
-        kv_cache: torch.Tensor,
-        attn_metadata: AttentionMetadata,
     ) -> torch.Tensor:
         q, _ = self.q_a_proj(hidden_states)
         q = self.q_a_layernorm(q)
@@ -170,7 +168,7 @@ class MiniCPM3Attention(nn.Module):
             v, [0, self.qk_head_dim - self.v_head_dim],
             value=0).view(-1, self.num_local_heads * self.qk_head_dim)
 
-        attn_output = self.attn(q, k, v, kv_cache, attn_metadata)
+        attn_output = self.attn(q, k, v)
         attn_output = attn_output.view(
             -1, self.num_local_heads,
             self.qk_head_dim)[..., :self.v_head_dim].reshape(