Error Codes
Smart Labeling
Error Type | Code |
---|---|
Smart labeling failed |
DL-W0401 |
DL-W0401
Cause:
-
Insufficient GPU memory. The GPU should have more than 6 GB of available memory.
-
Missing or outdated NVIDIA GPU driver.
Solution:
-
Free up GPU memory and then restart Mech-DLK to label again.
-
Use NVIDIA GPU driver version 472.50 or above.
Model Training
Error Type | Code |
---|---|
Training failed |
DL-E0101 |
DL-E0102 |
|
DL-E0103 |
|
DL-E0104 |
DL-E0101
Cause: Insufficient CUDA memory.
Solution: Instance Segmentation requires a GPU with a memory of at least 6 GB. Defect Segmentation requires a GPU with a memory of at least 4 GB. Free up GPU memory and restart training.
DL-E0102
Cause: Insufficient GPU memory for training.
Solution: Close some processes and restart training.
DL-E0103
Cause: Insufficient memory for training.
Solution: Close some processes and restart training.
DL-E0104
Cause:
-
The labels on the image are too small.
-
The ROI settings are inappropriate.
Solution:
-
Check if incorrect labels exist. Click here to view how to label images.
-
If all labels are correct, in the Training tab, click
, and appropriately increase the Input image size. -
Adjust the ROI. The ROI should select the object in the image and avoid irrelevant background.
Model Validation
Error Type | Code |
---|---|
Validation failed |
DL-W0201 |
DL-W0202 |
DL-W0201
Cause:
-
For models that use GPU for inference:
-
Insufficient GPU memory.
-
Outdated version of the NVIDIA driver.
-
-
For models that use CPU for inference: Insufficient CPU memory.
Solution:
-
For models that use GPU for inference:
-
Free up GPU memory and then restart Mech-DLK to restart validation.
-
Use NVIDIA GPU driver version 472.50 or above.
-
-
For models that use CPU for inference: Free up memory and then restart Mech-DLK to restart validation. In the installation directory of Mech-DLK, open the dl_sdk_log folder, and view the log from the corresponding time for troubleshooting.
DL-W0202
Cause:
-
Compatibility issues.
-
Insufficient GPU memory.
-
Outdated version of the NVIDIA driver.
Solution:
-
In the menu bar, click Settings, and enable Developer mode. In the Training tab, click
, and enable Finetune. Then, train the model again and validate the model. -
Free up GPU memory and then restart Mech-DLK to restart validation.
-
Use NVIDIA GPU driver version 472.50 or above.
Model Export
Error Type | Code |
---|---|
Model export failed |
DL-W0301 |
DL-W0302 |
DL-W0301
Cause:
-
The software installation path is too long.
-
Compatibility issues.
-
Python environment conflict.
Solution:
-
Change the installation path. Ensure the path length does not exceed 256 characters.
-
In the menu bar, click Settings, and enable Developer mode. In the Training tab, click
, and enable Finetune. Then, train the model again and export the model. -
Uninstall the Python environments and then try exporting again.
DL-W0302
Cause:
-
Compatibility issues.
-
Insufficient GPU memory.
-
Outdated version of the NVIDIA driver.
Solution:
-
In the menu bar, click Settings, and enable Developer mode. In the Training tab, click
, and enable Finetune. Then, train the model again and export the model. -
Free up GPU memory and then restart Mech-DLK to restart exporting.
-
Use NVIDIA GPU driver version 472.50 or above.
Operation Mode
Error Type | Code |
---|---|
Model loading failed |
DL-W0303 |
DL-W0303
Cause:
-
For models that use GPU for inference:
-
Insufficient GPU memory.
-
Outdated version of the NVIDIA driver.
-
-
For models that use CPU for inference: Insufficient CPU memory.
Solution:
-
For models that use GPU for inference:
-
Free up GPU memory and then restart Mech-DLK to restart validation.
-
Use NVIDIA GPU driver version 472.50 or above.
-
-
For models that use CPU for inference: Free up memory and then restart Mech-DLK to restart validation. In the installation directory of Mech-DLK, open the dl_sdk_log folder, and view the log from the corresponding time for troubleshooting.