Remove unused kwargs from model definitions (#13555)

vllm/model_executor/models/llama.py

@@ -22,13 +22,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Inference-only LLaMA model compatible with HuggingFace weights."""
-from typing import Any, Dict, Iterable, List, Optional, Set, Tuple, Type, Union
+from typing import Any, Dict, Iterable, Optional, Set, Tuple, Type, Union
 
 import torch
 from torch import nn
 from transformers import LlamaConfig
 
-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
@@ -197,13 +197,11 @@ class LlamaAttention(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.o_proj(attn_output)
         return output
 
@@ -268,8 +266,6 @@ class LlamaDecoderLayer(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
@@ -280,9 +276,7 @@ class LlamaDecoderLayer(nn.Module):
             hidden_states, residual = self.input_layernorm(
                 hidden_states, residual)
         hidden_states = self.self_attn(positions=positions,
-                                       hidden_states=hidden_states,
-                                       kv_cache=kv_cache,
-                                       attn_metadata=attn_metadata)
+                                       hidden_states=hidden_states)
 
         # Fully Connected
         hidden_states, residual = self.post_attention_layernorm(
@@ -347,8 +341,6 @@ class LlamaModel(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]:
@@ -363,11 +355,8 @@ class LlamaModel(nn.Module):
             hidden_states = intermediate_tensors["hidden_states"]
             residual = intermediate_tensors["residual"]
 
-        for i in range(self.start_layer, self.end_layer):
-            layer = self.layers[i]
-            hidden_states, residual = layer(positions, hidden_states,
-                                            kv_caches[i - self.start_layer],
-                                            attn_metadata, residual)
+        for layer in self.layers[self.start_layer:self.end_layer]:
+            hidden_states, residual = layer(positions, hidden_states, residual)
 
         if not get_pp_group().is_last_rank:
             return IntermediateTensors({
@@ -535,13 +524,10 @@ class LlamaForCausalLM(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.model(input_ids, positions, kv_caches,
-                                  attn_metadata, intermediate_tensors,
+        model_output = self.model(input_ids, positions, intermediate_tensors,
                                   inputs_embeds)
         return model_output