Remove unused kwargs from model definitions (#13555)

This commit is contained in:
Harry Mellor
2025-02-24 17:13:52 -08:00
committed by GitHub
parent f61528d46d
commit cdc1fa12eb
104 changed files with 436 additions and 1654 deletions
+6 -20
View File
@@ -22,13 +22,13 @@
# See the License for the specific language governing permissions and
# limitations under the License.
"""Inference-only Deepseek model."""
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.config import CacheConfig, VllmConfig
from vllm.distributed import (get_pp_group, get_tensor_model_parallel_rank,
get_tensor_model_parallel_world_size,
@@ -248,13 +248,11 @@ class DeepseekAttention(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
@@ -309,8 +307,6 @@ class DeepseekDecoderLayer(nn.Module):
self,
positions: torch.Tensor,
hidden_states: torch.Tensor,
kv_cache: torch.Tensor,
attn_metadata: AttentionMetadata,
residual: Optional[torch.Tensor],
) -> torch.Tensor:
# Self Attention
@@ -323,8 +319,6 @@ class DeepseekDecoderLayer(nn.Module):
hidden_states = self.self_attn(
positions=positions,
hidden_states=hidden_states,
kv_cache=kv_cache,
attn_metadata=attn_metadata,
)
# Fully Connected
@@ -370,8 +364,6 @@ class DeepseekModel(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]:
@@ -384,11 +376,8 @@ class DeepseekModel(nn.Module):
else:
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({
"hidden_states": hidden_states,
@@ -425,13 +414,10 @@ class DeepseekForCausalLM(nn.Module, 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]:
hidden_states = self.model(input_ids, positions, kv_caches,
attn_metadata, intermediate_tensors,
hidden_states = self.model(input_ids, positions, intermediate_tensors,
inputs_embeds)
return hidden_states