Neural machine translation with transformer decoded sentence is the same as input sentence

[jorcisai]

Issue Type

Bug

Source

source

Keras Version

3.6.0

Custom Code

Yes

OS Platform and Distribution

Linux Ubuntu 24.04

Python version

3.10.15

GPU model and memory

NVIDIA GeForce RTX 4060 8GB

Current Behavior?

I have noticed that the decoded sentence is the same as the input sentence (please see [1]).

Standalone code to reproduce the issue or tutorial link

https://colab.research.google.com/drive/1Y3_dgwqEQugOsoyth2Wu31YgORufzmsc?usp=sharing

Relevant log output

tf.Tensor(
[[ 34 358  31  42   0   0   0   0   0   0   0   0   0   0   0   0   0   0
    0   0]], shape=(1, 20), dtype=int64)
tf.Tensor(
[[  2 358  31  42   0   0   0   0   0   0   0   0   0   0   0   0   0   0
    0   0]], shape=(1, 20), dtype=int64)
I'm angry with her.
[start] estudiar las esta

[tomjelen]

This could be related to what got fixed in this pull request: #1997

[sachinprasadhs]

Hi, could you please try again, from the recent example output, it is working as expected, here is the link to the example guide: https://keras.io/examples/nlp/neural_machine_translation_with_transformer/

[github-actions]

This issue is stale because it has been open for 14 days with no activity. It will be closed if no further activity occurs. Thank you.

[github-actions]

This issue was closed because it has been inactive for 28 days. Please reopen if you'd like to work on this further.