gadagashwini

Results 47 comments of gadagashwini

Crashes with no error message but exit code -1073740791 (0xC0000409) on CUDA 11.7

Hi @aliencaocao, I am not seeing Crash message but training got stuck. Could you confirm the issue. Relevant log is below. Thank you! ``` 2022-09-19 10:40:19.283331: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1616] Created device...

tf.keras.layers.Layer.weights has inconsistent order in 'saved_model' format

HI @MathiasToftas, You need to use signatures under tf.function ( for graph mode). Please take a look at this reference [link ](https://www.tensorflow.org/lite/guide/signatures#:~:text=%40tf.function%28input_signature%3D%5Btf.TensorSpec%28shape%3D%5BNone%5D%2C%20dtype%3Dtf.float32%29%5D%29)for more information. Thank you!

error in `tf.train.import_meta_graph()`

Hi @wwdok, Issue with input data shape. in your case, x/y has shape [512, 0] while x/y has shape [512]. they are not compatible. Please try to reshape input data...

No convenient way to apply class weights for a segmentation model

@AngledLuffa, Please follow the workarounds mentioned on similar thread [#3653](https://github.com/keras-team/keras/issues/3653). And also refer this [issue](https://github.com/tensorflow/tensorflow/issues/48714). Thanks!

No convenient way to apply class weights for a segmentation model

Hi @AngledLuffa, Could you try the workaround mentioned in the comment and let us know if that solves your problem, Thank you!

tf.keras.utils.split_dataset does not seem to work

Hi @frigeriomtt, Crash can be due to large dataset. Share the dataset. Thank you!

tf.ragged.constant does not detect dense dimensions

Was able to reproduce the issue with Tf-nightly 2.10.0-dev20220726 .Please find the gist [here](https://colab.research.google.com/drive/15uSipleFhnAXJC9kP_pTydai9G9XlZv2?usp=sharing). Thank you!