Remove unused kwargs from model definitions (#13555)

vllm/model_executor/models/qwen.py

@@ -6,13 +6,13 @@
 # LICENSE: https://huggingface.co/Qwen/Qwen-7B/blob/main/LICENSE
 """Inference-only QWen 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 transformers import PretrainedConfig
 
-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
@@ -124,13 +124,11 @@ class QWenAttention(nn.Module):
         self,
         positions: torch.Tensor,
         hidden_states: torch.Tensor,
-        kv_cache: torch.Tensor,
-        attn_metadata: AttentionMetadata,
     ) -> torch.Tensor:
         qkv, _ = self.c_attn(hidden_states)
         q, k, v = qkv.chunk(chunks=3, 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.c_proj(attn_output)
         return output
 
@@ -168,8 +166,6 @@ class QWenBlock(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
@@ -181,8 +177,6 @@ class QWenBlock(nn.Module):
         hidden_states = self.attn(
             positions=positions,
             hidden_states=hidden_states,
-            kv_cache=kv_cache,
-            attn_metadata=attn_metadata,
         )
 
         # Fully Connected
@@ -225,8 +219,6 @@ class QWenModel(nn.Module):
         self,
         input_ids: 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]:
@@ -241,13 +233,10 @@ class QWenModel(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,
             )
         if not get_pp_group().is_last_rank:
@@ -373,12 +362,9 @@ class QWenLMHeadModel(QWenBaseModel, SupportsPP, SupportsLoRA):
         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]:
-        hidden_states = self.transformer(input_ids, positions, kv_caches,
-                                         attn_metadata, intermediate_tensors,
-                                         inputs_embeds)
+        hidden_states = self.transformer(input_ids, positions,
+                                         intermediate_tensors, inputs_embeds)
         return hidden_states