Skip to content

Fix protobuf serialization failure in ort_fusions ort_run for large models #2955

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
justinchuby merged 3 commits into from
Jul 9, 2026
+12 −5
Merged

Fix protobuf serialization failure in ort_fusions ort_run for large models #2955

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
2f27cc7
Initial plan
Copilot Jul 8, 2026
f4c1f3a
Fix ort_run serialization failure by saving model with external data
Copilot Jul 8, 2026
c0b2527
Merge branch 'main' into copilot/investigate-fix-ci-issues
justinchuby Jul 8, 2026
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
17 changes: 12 additions & 5 deletions onnxscript/rewriter/ort_fusions/_test_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,9 @@
# Licensed under the MIT License.
from __future__ import annotations

import os
import tempfile

import numpy as np
import onnx_ir as ir
import onnxruntime
Expand All @@ -12,13 +15,17 @@

def ort_run(model_name: str, model, inputs):
providers = ["CPUExecutionProvider"]
model_proto = ir.serde.serialize_model(model)
options = onnxruntime.SessionOptions()
options.graph_optimization_level = onnxruntime.GraphOptimizationLevel.ORT_DISABLE_ALL
session = onnxruntime.InferenceSession(
model_proto.SerializeToString(), options, providers=providers
)
return session.run(None, inputs)
# Save the model to a temporary file and load it from disk. This uses external data
# to store tensors so that large models are not limited by the 2GB protobuf
# serialization limit, which can otherwise raise
# ``google.protobuf.message.EncodeError: Failed to serialize proto``.
with tempfile.TemporaryDirectory() as temp_dir:
model_path = os.path.join(temp_dir, f"{model_name}.onnx")
ir.save(model, model_path, external_data=f"{model_name}.onnx.data")
session = onnxruntime.InferenceSession(model_path, options, providers=providers)
return session.run(None, inputs)


def assert_allclose(outputs, expected_outputs, rtol=1e-3, atol=1e-3):
Expand Down
Loading