Skip to content

[Graph Optimization] SOT+CUDAGraph support ERNIE4.5T VL 28B / 424B #4645

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 7 commits into from
Oct 31, 2025
Merged

[Graph Optimization] SOT+CUDAGraph support ERNIE4.5T VL 28B / 424B #4645

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
95d3c2d
45TVL support sot+CUDAGraph
DrRyanHuang Oct 29, 2025
ce2aa67
mv unitest from ce_deploy 2 e2e
DrRyanHuang Oct 29, 2025
1c4a9e1
Merge branch 'develop' into 45t_vl_support_sot_cudagraph
DrRyanHuang Oct 29, 2025
b2f6e70
add test_EB_VL_Lite_sot_serving
DrRyanHuang Oct 29, 2025
5e19d19
rm useless line
DrRyanHuang Oct 29, 2025
15d7d91
add openai_client
DrRyanHuang Oct 30, 2025
ff6d671
fix unitest && reduce computing resources
DrRyanHuang Oct 30, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion benchmarks/benchmark_serving.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -982,7 +982,7 @@ def main(args: argparse.Namespace):
if args.result_dir:
file_name = os.path.join(args.result_dir, file_name)
with open(file_name, "w", encoding="utf-8") as outfile:
json.dump(result_json, outfile)
json.dump(result_json, outfile, ensure_ascii=False)
save_to_pytorch_benchmark_format(args, result_json, file_name)


Expand Down
6 changes: 3 additions & 3 deletions fastdeploy/model_executor/models/ernie4_5_vl/ernie4_5_vl_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -277,7 +277,7 @@ def load_state_dict(self, state_dict):
def forward(self, hidden_states: paddle.Tensor, vl_moe_meta: VLMoEMeta):
if self.num_shared_experts > 0:
shared_experts_out = self.shared_experts(hidden_states)
hidden_states, vl_moe_meta.text_input, vl_moe_meta.image_input = text_image_gather_scatter(
hidden_states, text_input, image_input = text_image_gather_scatter(
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

这里和xpu那个修改会冲突吗

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

XPU 的 #4636 合入后，这个再合入就没问题

hidden_states,
vl_moe_meta.text_input,
vl_moe_meta.image_input,
Expand All @@ -286,8 +286,8 @@ def forward(self, hidden_states: paddle.Tensor, vl_moe_meta: VLMoEMeta):
vl_moe_meta.image_index,
True,
)
text_out = self.text_fused_moe(vl_moe_meta.text_input)
image_out = self.image_fused_moe(vl_moe_meta.image_input)
text_out = self.text_fused_moe(text_input)
image_out = self.image_fused_moe(image_input)
hidden_states, _, _ = text_image_gather_scatter(
hidden_states,
text_out,
Expand Down
Loading
Loading