Remove unused kwargs from model definitions (#13555)
This commit is contained in:
@@ -22,7 +22,7 @@
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
"""Inference-only Mixtral model."""
|
||||
from typing import Iterable, List, Optional, Set, Tuple, Union
|
||||
from typing import Iterable, Optional, Set, Tuple, Union
|
||||
|
||||
import numpy as np
|
||||
import torch
|
||||
@@ -30,7 +30,7 @@ import torch.nn.functional as F
|
||||
from torch import nn
|
||||
from transformers import MixtralConfig
|
||||
|
||||
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,
|
||||
@@ -229,13 +229,11 @@ class MixtralAttention(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
|
||||
|
||||
@@ -274,8 +272,6 @@ class MixtralDecoderLayer(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
|
||||
@@ -288,8 +284,6 @@ class MixtralDecoderLayer(nn.Module):
|
||||
hidden_states = self.self_attn(
|
||||
positions=positions,
|
||||
hidden_states=hidden_states,
|
||||
kv_cache=kv_cache,
|
||||
attn_metadata=attn_metadata,
|
||||
)
|
||||
|
||||
# Fully Connected
|
||||
@@ -333,8 +327,6 @@ class MixtralModel(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]:
|
||||
@@ -348,11 +340,8 @@ class MixtralModel(nn.Module):
|
||||
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]
|
||||
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,
|
||||
@@ -390,13 +379,10 @@ class MixtralForCausalLM(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
|
||||
|
||||
|
||||
Reference in New Issue
Block a user