-
Notifications
You must be signed in to change notification settings - Fork 140
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.
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
RuntimeError with tensor dimension using rotrain #659
Comments
On 24/11/08 06:29AM, Svetlana Yatsyk wrote:
Could you please help me understand why this tensor dimension mismatch
error is persisting on the university server, even though the same
adjustments worked in Google Colab? I would appreciate any guidance on
overcoming these issues.
I'd need a stack trace to say more. The number of points in the baseline
shouldn't have any impact on the training process baring any bugs.
|
My Colab training just stopped with same old error To the data : https://msia.escriptorium.fr/media/users/334/export_doc2968_biblioteca_laurenziana_plutei_28_pagexml_202411081332.zip And here are the university server tracebacks: For batch = 512
And for batch = 1000 (I believe that nothing else changed):
|
Hello! I'm trying to train a reading order model. I'm running the same training script on both Google Colab and my university server. The kraken version is identical on both systems (5.2.9). Here is the script I'm using:
While the training starts successfully in Google Colab, it fails on the university server with the following error:
RuntimeError: stack expects each tensor to be equal size, but got [18] at entry 0 and [16] at entry 8
I encountered a similar issue on Colab initially, and I thought it was due to inconsistencies in how baselines were defined (some baselines were defined with 2, some with 3, and some with 4 points). To resolve this, I filtered out all baselines except those defined by 2 points, and after that, the training started successfully in Colab. However, the same approach did not resolve the issue on the university server. When I try increasing the batch size on the server, I encounter a different error:
RuntimeError: Trying to resize storage that is not resizable
Could you please help me understand why this tensor dimension mismatch error is persisting on the university server, even though the same adjustments worked in Google Colab? I would appreciate any guidance on overcoming these issues.
The text was updated successfully, but these errors were encountered: