Remove unused kwargs from model definitions (#13555)

vllm/model_executor/models/internlm2_ve.py

@@ -1,12 +1,11 @@
 # SPDX-License-Identifier: Apache-2.0
 
-from typing import List, Optional, Tuple, Union
+from typing import Optional, Tuple, Union
 
 import torch
 from torch import nn
 from transformers import PretrainedConfig
 
-from vllm.attention import AttentionMetadata
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group
 from vllm.model_executor.layers.layernorm import RMSNorm
@@ -65,8 +64,6 @@ class InternLM2VEDecoderLayer(nn.Module):
         self,
         positions: torch.Tensor,
         hidden_states: torch.Tensor,
-        kv_cache: torch.Tensor,
-        attn_metadata: AttentionMetadata,
         residual: Optional[torch.Tensor],
         visual_token_mask: Optional[torch.Tensor] = None,
     ) -> Tuple[torch.Tensor, torch.Tensor]:
@@ -80,8 +77,6 @@ class InternLM2VEDecoderLayer(nn.Module):
         hidden_states = self.attention(
             positions=positions,
             hidden_states=hidden_states,
-            kv_cache=kv_cache,
-            attn_metadata=attn_metadata,
         )
 
         # Fully Connected
@@ -113,8 +108,6 @@ class InternLM2VEModel(InternLM2Model):
         self,
         input_ids: torch.Tensor,
         positions: torch.Tensor,
-        kv_caches: List[torch.Tensor],
-        attn_metadata: AttentionMetadata,
         intermediate_tensors: Optional[IntermediateTensors] = None,
         inputs_embeds: Optional[torch.Tensor] = None,
         visual_token_mask: Optional[torch.Tensor] = None,
@@ -129,13 +122,10 @@ class InternLM2VEModel(InternLM2Model):
             assert intermediate_tensors is not None
             hidden_states = intermediate_tensors["hidden_states"]
             residual = intermediate_tensors["residual"]
-        for i in range(self.start_layer, self.end_layer):
-            layer = self.layers[i]
+        for layer in self.layers[self.start_layer:self.end_layer]:
             hidden_states, residual = layer(
                 positions,
                 hidden_states,
-                kv_caches[i - self.start_layer],
-                attn_metadata,
                 residual,
                 visual_token_mask=visual_token_mask,
             )