apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 08:37:06 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[Executor] Move forward_meta.py to fastdeploy/model_executor (#2774)

Browse Source 
* Use PEP 563 in attention.py and fix conflict

* merge commit

* Change what was left out last time
This commit is contained in:
littledgg
2025-07-10 20:36:51 +08:00
committed by GitHub
parent 8c660a0dfb
commit 59071268b6
27 changed files with 53 additions and 55 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
fastdeploy/model_executor/models/deepseek_v3.py
View File

@@ -40,7 +40,7 @@ from fastdeploy.model_executor.layers.rotary_embedding import \
DeepseekScalingRotaryEmbedding
from fastdeploy.model_executor.models.model_base import ModelForCasualLM
from fastdeploy.platforms import current_platform
from fastdeploy.worker.forward_meta import ForwardMeta
from fastdeploy.model_executor.forward_meta import ForwardMeta
if current_platform.is_cuda():
from fastdeploy.model_executor.ops.gpu import \

2
fastdeploy/model_executor/models/ernie4_5_moe.py
View File

@@ -41,7 +41,7 @@ from fastdeploy.model_executor.models.tp_utils import TensorSplitMode as tsm
from fastdeploy.model_executor.models.utils import \
LayerIdPlaceholder as layerid
from fastdeploy.model_executor.models.utils import WeightMeta
from fastdeploy.worker.forward_meta import ForwardMeta
from fastdeploy.model_executor.forward_meta import ForwardMeta
class Ernie4_5_MLP(nn.Layer):

2
fastdeploy/model_executor/models/ernie4_5_mtp.py
View File

@@ -30,7 +30,7 @@ from fastdeploy.model_executor.layers.mtp_linear import ParallelEHProjection
from fastdeploy.model_executor.layers.normalization import RMSNorm
from fastdeploy.model_executor.models.ernie4_5_moe import Ernie4_5_DecoderLayer
from fastdeploy.model_executor.models.model_base import ModelForCasualLM
from fastdeploy.worker.forward_meta import ForwardMeta
from fastdeploy.model_executor.forward_meta import ForwardMeta
class Ernie4_5_MTPPretrainedModel(PretrainedModel):

2
fastdeploy/model_executor/models/ernie4_5_vl/ernie4_5_vl_moe.py
View File

@@ -44,7 +44,7 @@ if current_platform.is_cuda() and not current_platform.is_dcu():
text_image_gather_scatter,
text_image_index_out)
from fastdeploy.worker.forward_meta import ForwardMeta
from fastdeploy.model_executor.forward_meta import ForwardMeta
class Ernie4_5_VLMLP(Ernie4_5_MLP):

2
fastdeploy/model_executor/models/qwen2.py
View File

@@ -34,7 +34,7 @@ from fastdeploy.model_executor.layers.linear import (
from fastdeploy.model_executor.layers.lm_head import ParallelLMHead
from fastdeploy.model_executor.layers.normalization import RMSNorm
from fastdeploy.model_executor.models.model_base import ModelForCasualLM
from fastdeploy.worker.forward_meta import ForwardMeta
from fastdeploy.model_executor.forward_meta import ForwardMeta
class Qwen2MLP(nn.Layer):

2
fastdeploy/model_executor/models/qwen3.py
View File

@@ -34,7 +34,7 @@ from fastdeploy.model_executor.layers.lm_head import ParallelLMHead
from fastdeploy.model_executor.layers.normalization import RMSNorm
from fastdeploy.model_executor.models.model_base import ModelForCasualLM
from fastdeploy.model_executor.models.qwen2 import Qwen2DecoderLayer, Qwen2MLP
from fastdeploy.worker.forward_meta import ForwardMeta
from fastdeploy.model_executor.forward_meta import ForwardMeta
class Qwen3MLP(Qwen2MLP):

2
fastdeploy/model_executor/models/qwen3moe.py
View File

@@ -35,7 +35,7 @@ from fastdeploy.model_executor.layers.lm_head import ParallelLMHead
from fastdeploy.model_executor.layers.moe.moe import FusedMoE
from fastdeploy.model_executor.layers.normalization import RMSNorm
from fastdeploy.model_executor.models.model_base import ModelForCasualLM
from fastdeploy.worker.forward_meta import ForwardMeta
from fastdeploy.model_executor.forward_meta import ForwardMeta
class Qwen3MLP(nn.Layer):