Remove unused kwargs from model definitions (#13555)

vllm/model_executor/models/exaone.py

@@ -24,12 +24,12 @@
 # limitations under the License.
 """Inference-only Exaone model compatible with HuggingFace weights."""
 
-from typing import Any, Dict, Iterable, List, Optional, Set, Tuple, Union
+from typing import Any, Dict, Iterable, Optional, Set, Tuple, Union
 
 import torch
 from torch import nn
 
-from vllm.attention import Attention, AttentionMetadata
+from vllm.attention import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size
@@ -179,13 +179,11 @@ class ExaoneAttention(nn.Module):
         self,
         positions: torch.Tensor,
         hidden_states: torch.Tensor,
-        kv_cache: torch.Tensor,
-        attn_metadata: AttentionMetadata,
     ) -> torch.Tensor:
         qkv, _ = self.qkv_proj(hidden_states)
         q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)
         q, k = self.rotary_emb(positions, q, k)
-        attn_output = self.attn(q, k, v, kv_cache, attn_metadata)
+        attn_output = self.attn(q, k, v)
         output, _ = self.out_proj(attn_output)
         return output
 
@@ -225,14 +223,10 @@ class ExaoneBlockAttention(nn.Module):
         self,
         positions: torch.Tensor,
         hidden_states: torch.Tensor,
-        kv_cache: torch.Tensor,
-        attn_metadata: AttentionMetadata,
     ) -> torch.Tensor:
         return self.attention(
             positions=positions,
             hidden_states=hidden_states,
-            kv_cache=kv_cache,
-            attn_metadata=attn_metadata,
         )
 
 
@@ -288,8 +282,6 @@ class ExaoneDecoderLayer(nn.Module):
         self,
         positions: torch.Tensor,
         hidden_states: torch.Tensor,
-        kv_cache: torch.Tensor,
-        attn_metadata: AttentionMetadata,
         residual: Optional[torch.Tensor],
     ) -> Tuple[torch.Tensor, torch.Tensor]:
         # Self Attention
@@ -301,8 +293,6 @@ class ExaoneDecoderLayer(nn.Module):
         hidden_states = self.attn(
             positions=positions,
             hidden_states=hidden_states,
-            kv_cache=kv_cache,
-            attn_metadata=attn_metadata,
         )
 
         # Fully Connected
@@ -365,8 +355,6 @@ class ExaoneModel(nn.Module):
         self,
         input_ids: Optional[torch.Tensor],
         positions: torch.Tensor,
-        kv_caches: List[torch.Tensor],
-        attn_metadata: AttentionMetadata,
         intermediate_tensors: Optional[IntermediateTensors],
         inputs_embeds: Optional[torch.Tensor] = None,
     ) -> Union[torch.Tensor, IntermediateTensors]:
@@ -381,13 +369,10 @@ class ExaoneModel(nn.Module):
             hidden_states = intermediate_tensors["hidden_states"]
             residual = intermediate_tensors["residual"]
 
-        for i in range(self.start_layer, self.end_layer):
-            layer = self.h[i]
+        for layer in self.h[self.start_layer:self.end_layer]:
             hidden_states, residual = layer(
                 positions,
                 hidden_states,
-                kv_caches[i - self.start_layer],
-                attn_metadata,
                 residual,
             )
 
@@ -471,14 +456,11 @@ class ExaoneForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
         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,
     ) -> Union[torch.Tensor, IntermediateTensors]:
-        model_output = self.transformer(input_ids, positions, kv_caches,
-                                        attn_metadata, intermediate_tensors,
-                                        inputs_embeds)
+        model_output = self.transformer(input_ids, positions,
+                                        intermediate_tensors, inputs_embeds)
         return model_output
 
     def compute_logits(