[go: nahoru, domu]
Skip to content
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

Jump to bottom

Overflow in tf.compat.v1.nn.ctc_loss and tf.compat.v1.nn.ctc_loss_v2 when there are too large values in labels. #65864

Open
xl1043237213 opened this issue Apr 16, 2024 · 1 comment
Open

Overflow in tf.compat.v1.nn.ctc_loss and tf.compat.v1.nn.ctc_loss_v2 when there are too large values in labels. #65864

xl1043237213 opened this issue Apr 16, 2024 · 1 comment
Assignees
@SuryanarayanaY
Labels
comp:ops OPs related issues stat:awaiting tensorflower Status - Awaiting response from tensorflower TF 2.16 type:bug Bug

Comments

@xl1043237213
Copy link
xl1043237213 commented Apr 16, 2024 •
edited
Loading

Issue type

Bug

Have you reproduced the bug with TensorFlow Nightly?

Yes

Source

binary

TensorFlow version

tf 2.15

Custom code

No

OS platform and distribution

Linux Ubuntu 20.04

Mobile device

No response

Python version

3.10

Bazel version

No response

GCC/compiler version

No response

CUDA/cuDNN version

No response

GPU model and memory

No response

Current behavior?

Overflow in tf.compat.v1.nn.ctc_loss and tf.compat.v1.nn.ctc_loss_v2 when the value of labels is too large. See the colab link below for details.

Standalone code to reproduce the issue

https://colab.research.google.com/drive/1zKblem8QIsfEA973gnXrY6NDLch6ysoa?usp=sharing

Relevant log output

Non-OK-status: InitDims(dim_sizes) status: INVALID_ARGUMENT: Encountered overflow when multiplying 2 with 9143586203607084488, result: -159571666495382640
@google-ml-butler google-ml-butler bot added the type:bug Bug label Apr 16, 2024
@xl1043237213 xl1043237213 changed the title Overflow in tf.compat.v1.nn.ctc_loss when the value of labels is too large. Overflow in tf.compat.v1.nn.ctc_loss when there are too large values in labels. Apr 16, 2024
@xl1043237213 xl1043237213 changed the title Overflow in tf.compat.v1.nn.ctc_loss when there are too large values in labels. Overflow in tf.compat.v1.nn.ctc_loss and tf.compat.v1.nn.ctc_loss_v2 when there are too large values in labels. Apr 16, 2024
@SuryanarayanaY
Copy link
Collaborator

Hi @xl1043237213 ,

I have replicated the reported behaviour and attached screenshot below. The overflow warning is expected but it follows a crash which might needs fix.

Screenshot 2024-04-17 at 11 38 46 AM

@SuryanarayanaY SuryanarayanaY added comp:ops OPs related issues TF 2.16 labels Apr 17, 2024
@SuryanarayanaY SuryanarayanaY added the stat:awaiting tensorflower Status - Awaiting response from tensorflower label Apr 25, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@SuryanarayanaY SuryanarayanaY

Labels
comp:ops OPs related issues stat:awaiting tensorflower Status - Awaiting response from tensorflower TF 2.16 type:bug Bug
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@xl1043237213 @SuryanarayanaY