Skip to content

Fix failing yolo26 gpu integration tests #1956

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
PawelPeczek-Roboflow merged 4 commits into from
Jan 30, 2026
Merged

Fix failing yolo26 gpu integration tests #1956

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
415e0c9
add back CUDAExecutionProvider
mkaic Jan 29, 2026
7019fec
update tests to handle divergent CPU/GPU values
mkaic Jan 30, 2026
0a55212
i think it is finally actually maybe fixed
mkaic Jan 30, 2026
883c053
generate GPU values on T4 instead of L4
mkaic Jan 30, 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
24 changes: 12 additions & 12 deletions ...dels/tests/integration_tests/models/test_yolo26_instance_segmentation_predictions_onnx.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ def test_onnx_package_stretch_static_bs_numpy(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_stretch_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model(snake_image_numpy)
Expand Down Expand Up @@ -62,7 +62,7 @@ def test_onnx_package_stretch_static_bs_batch_numpy(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_stretch_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model([snake_image_numpy, snake_image_numpy])
Expand Down Expand Up @@ -128,7 +128,7 @@ def test_onnx_package_stretch_static_bs_torch(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_stretch_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model(snake_image_torch)
Expand Down Expand Up @@ -172,7 +172,7 @@ def test_onnx_package_stretch_dynamic_bs_numpy(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_stretch_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model(snake_image_numpy)
Expand Down Expand Up @@ -216,7 +216,7 @@ def test_onnx_package_stretch_dynamic_bs_batch_numpy(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_stretch_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model([snake_image_numpy, snake_image_numpy])
Expand Down Expand Up @@ -282,7 +282,7 @@ def test_onnx_package_stretch_dynamic_bs_torch(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_stretch_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model(snake_image_torch)
Expand Down Expand Up @@ -326,7 +326,7 @@ def test_onnx_package_letterbox_static_bs_numpy(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_letterbox_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model(snake_image_numpy)
Expand Down Expand Up @@ -370,7 +370,7 @@ def test_onnx_package_letterbox_static_bs_batch_numpy(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_letterbox_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model([snake_image_numpy, snake_image_numpy])
Expand Down Expand Up @@ -436,7 +436,7 @@ def test_onnx_package_letterbox_static_bs_torch(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_letterbox_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model(snake_image_torch)
Expand Down Expand Up @@ -480,7 +480,7 @@ def test_onnx_package_letterbox_dynamic_bs_numpy(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_letterbox_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model(snake_image_numpy)
Expand Down Expand Up @@ -524,7 +524,7 @@ def test_onnx_package_letterbox_dynamic_bs_batch_numpy(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_letterbox_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model([snake_image_numpy, snake_image_numpy])
Expand Down Expand Up @@ -590,7 +590,7 @@ def test_onnx_package_letterbox_dynamic_bs_torch(

model = YOLO26ForInstanceSegmentationOnnx.from_pretrained(
model_name_or_path=yolo26n_seg_snakes_letterbox_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
)

predictions = model(snake_image_torch)
Expand Down
104 changes: 52 additions & 52 deletions ...models/tests/integration_tests/models/test_yolo26_keypoints_detection_predictions_onnx.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,9 +2,16 @@
import pytest
import torch

from inference_models.configuration import DEFAULT_DEVICE

COORD_TOLERANCE = 2
CONF_TOLERANCE = 0.01

IS_GPU = str(DEFAULT_DEVICE).startswith("cuda")
ONNX_EXECUTION_PROVIDERS = (
["CUDAExecutionProvider", "CPUExecutionProvider"] if IS_GPU else ["CPUExecutionProvider"]
)

LETTERBOX_EXPECTED_KP_XY_1 = [
[617, 297],
[0, 0],
Expand Down Expand Up @@ -221,6 +228,9 @@
0.0,
]

CPU_LETTERBOX_BOX_XYXY = [[-33, 246, 1732, 1076], [-8, 252, 1798, 1076]]
CPU_LETTERBOX_BOX_CONF = [0.9278, 0.4428]


@pytest.mark.slow
@pytest.mark.onnx_extras
Expand All @@ -234,7 +244,7 @@ def test_yolo26n_pose_onnx_static_letterbox_numpy(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_letterbox_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model(basketball_image_numpy)
Expand All @@ -252,17 +262,15 @@ def test_yolo26n_pose_onnx_static_letterbox_numpy(
torch.tensor([LETTERBOX_EXPECTED_KP_CONF_1, LETTERBOX_EXPECTED_KP_CONF_2]),
atol=CONF_TOLERANCE,
)

expected_box_xyxy = torch.tensor(CPU_LETTERBOX_BOX_XYXY, dtype=torch.int32)
expected_box_conf = torch.tensor(CPU_LETTERBOX_BOX_CONF)

assert torch.allclose(
predictions[1][0].xyxy.cpu(),
torch.tensor(
[[-33, 246, 1732, 1076], [-8, 252, 1798, 1076]], dtype=torch.int32
),
atol=COORD_TOLERANCE,
predictions[1][0].xyxy.cpu(), expected_box_xyxy, atol=COORD_TOLERANCE
)
assert torch.allclose(
predictions[1][0].confidence.cpu(),
torch.tensor([0.9278, 0.4428]),
atol=CONF_TOLERANCE,
predictions[1][0].confidence.cpu(), expected_box_conf, atol=CONF_TOLERANCE
)


Expand All @@ -278,7 +286,7 @@ def test_yolo26n_pose_onnx_static_letterbox_batch_numpy(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_letterbox_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model([basketball_image_numpy, basketball_image_numpy])
Expand All @@ -290,10 +298,9 @@ def test_yolo26n_pose_onnx_static_letterbox_batch_numpy(
expected_kp_conf = torch.tensor(
[LETTERBOX_EXPECTED_KP_CONF_1, LETTERBOX_EXPECTED_KP_CONF_2]
)
expected_box_xyxy = torch.tensor(
[[-33, 246, 1732, 1076], [-8, 252, 1798, 1076]], dtype=torch.int32
)
expected_box_conf = torch.tensor([0.9278, 0.4428])

expected_box_xyxy = torch.tensor(CPU_LETTERBOX_BOX_XYXY, dtype=torch.int32)
expected_box_conf = torch.tensor(CPU_LETTERBOX_BOX_CONF)

for i in range(2):
assert torch.allclose(
Expand Down Expand Up @@ -322,7 +329,7 @@ def test_yolo26n_pose_onnx_static_letterbox_torch(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_letterbox_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model(basketball_image_torch)
Expand All @@ -340,17 +347,15 @@ def test_yolo26n_pose_onnx_static_letterbox_torch(
torch.tensor([LETTERBOX_EXPECTED_KP_CONF_1, LETTERBOX_EXPECTED_KP_CONF_2]),
atol=CONF_TOLERANCE,
)

expected_box_xyxy = torch.tensor(CPU_LETTERBOX_BOX_XYXY, dtype=torch.int32)
expected_box_conf = torch.tensor(CPU_LETTERBOX_BOX_CONF)

assert torch.allclose(
predictions[1][0].xyxy.cpu(),
torch.tensor(
[[-33, 246, 1732, 1076], [-8, 252, 1798, 1076]], dtype=torch.int32
),
atol=COORD_TOLERANCE,
predictions[1][0].xyxy.cpu(), expected_box_xyxy, atol=COORD_TOLERANCE
)
assert torch.allclose(
predictions[1][0].confidence.cpu(),
torch.tensor([0.9278, 0.4428]),
atol=CONF_TOLERANCE,
predictions[1][0].confidence.cpu(), expected_box_conf, atol=CONF_TOLERANCE
)


Expand All @@ -366,7 +371,7 @@ def test_yolo26n_pose_onnx_dynamic_letterbox_numpy(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_letterbox_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model(basketball_image_numpy)
Expand All @@ -384,17 +389,15 @@ def test_yolo26n_pose_onnx_dynamic_letterbox_numpy(
torch.tensor([LETTERBOX_EXPECTED_KP_CONF_1, LETTERBOX_EXPECTED_KP_CONF_2]),
atol=CONF_TOLERANCE,
)

expected_box_xyxy = torch.tensor(CPU_LETTERBOX_BOX_XYXY, dtype=torch.int32)
expected_box_conf = torch.tensor(CPU_LETTERBOX_BOX_CONF)

assert torch.allclose(
predictions[1][0].xyxy.cpu(),
torch.tensor(
[[-33, 246, 1732, 1076], [-8, 252, 1798, 1076]], dtype=torch.int32
),
atol=COORD_TOLERANCE,
predictions[1][0].xyxy.cpu(), expected_box_xyxy, atol=COORD_TOLERANCE
)
assert torch.allclose(
predictions[1][0].confidence.cpu(),
torch.tensor([0.9278, 0.4428]),
atol=CONF_TOLERANCE,
predictions[1][0].confidence.cpu(), expected_box_conf, atol=CONF_TOLERANCE
)


Expand All @@ -410,7 +413,7 @@ def test_yolo26n_pose_onnx_dynamic_letterbox_batch_numpy(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_letterbox_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model([basketball_image_numpy, basketball_image_numpy])
Expand All @@ -422,10 +425,9 @@ def test_yolo26n_pose_onnx_dynamic_letterbox_batch_numpy(
expected_kp_conf = torch.tensor(
[LETTERBOX_EXPECTED_KP_CONF_1, LETTERBOX_EXPECTED_KP_CONF_2]
)
expected_box_xyxy = torch.tensor(
[[-33, 246, 1732, 1076], [-8, 252, 1798, 1076]], dtype=torch.int32
)
expected_box_conf = torch.tensor([0.9278, 0.4428])

expected_box_xyxy = torch.tensor(CPU_LETTERBOX_BOX_XYXY, dtype=torch.int32)
expected_box_conf = torch.tensor(CPU_LETTERBOX_BOX_CONF)

for i in range(2):
assert torch.allclose(
Expand Down Expand Up @@ -454,7 +456,7 @@ def test_yolo26n_pose_onnx_dynamic_letterbox_torch(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_letterbox_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model(basketball_image_torch)
Expand All @@ -472,17 +474,15 @@ def test_yolo26n_pose_onnx_dynamic_letterbox_torch(
torch.tensor([LETTERBOX_EXPECTED_KP_CONF_1, LETTERBOX_EXPECTED_KP_CONF_2]),
atol=CONF_TOLERANCE,
)

expected_box_xyxy = torch.tensor(CPU_LETTERBOX_BOX_XYXY, dtype=torch.int32)
expected_box_conf = torch.tensor(CPU_LETTERBOX_BOX_CONF)

assert torch.allclose(
predictions[1][0].xyxy.cpu(),
torch.tensor(
[[-33, 246, 1732, 1076], [-8, 252, 1798, 1076]], dtype=torch.int32
),
atol=COORD_TOLERANCE,
predictions[1][0].xyxy.cpu(), expected_box_xyxy, atol=COORD_TOLERANCE
)
assert torch.allclose(
predictions[1][0].confidence.cpu(),
torch.tensor([0.9278, 0.4428]),
atol=CONF_TOLERANCE,
predictions[1][0].confidence.cpu(), expected_box_conf, atol=CONF_TOLERANCE
)


Expand All @@ -498,7 +498,7 @@ def test_yolo26n_pose_onnx_static_stretch_numpy(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_stretch_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model(basketball_image_numpy)
Expand Down Expand Up @@ -537,7 +537,7 @@ def test_yolo26n_pose_onnx_static_stretch_batch_numpy(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_stretch_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model([basketball_image_numpy, basketball_image_numpy])
Expand Down Expand Up @@ -574,7 +574,7 @@ def test_yolo26n_pose_onnx_static_stretch_torch(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_stretch_onnx_static_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model(basketball_image_torch)
Expand Down Expand Up @@ -613,7 +613,7 @@ def test_yolo26n_pose_onnx_dynamic_stretch_numpy(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_stretch_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model(basketball_image_numpy)
Expand Down Expand Up @@ -652,7 +652,7 @@ def test_yolo26n_pose_onnx_dynamic_stretch_batch_numpy(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_stretch_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model([basketball_image_numpy, basketball_image_numpy])
Expand Down Expand Up @@ -689,7 +689,7 @@ def test_yolo26n_pose_onnx_dynamic_stretch_torch(

model = YOLO26ForKeyPointsDetectionOnnx.from_pretrained(
model_name_or_path=yolo26n_pose_basketball_stretch_onnx_dynamic_package,
onnx_execution_providers=["CPUExecutionProvider"],
onnx_execution_providers=ONNX_EXECUTION_PROVIDERS,
)

predictions = model(basketball_image_torch)
Expand Down
Loading