keras-io
keras-io copied to clipboard
GPT text generation with KerasNLP Example error
Hello, in the GPT text generation example with KerasNLP (https://keras.io/examples/nlp/text_generation_gpt/)
when executing the code associated with # Train tokenizer vocabulary the following error is thrown:
AttributeError: module 'keras_nlp.tokenizers' has no attribute 'compute_word_piece_vocabulary'
I have tried with several versions of keras-nlp but I have not fixed the error.
@ygModesto Actually the error is caused because the latest release of KerasNLP doesn't yet have the compute_word_piece_vocabulary
function. If you want to run it you can install KerasNLP from github directly using this syntax -
pip install git+https://github.com/user/repo.git@branch
@Alpes-ai Hey that was just a placeholder should've been more clearer :( Instead of that link use the link from the KerasNLP repository with the master branch, hope that sorts it!
Same here am getting the error when I pip install []pip install git+https://github.com/user/repo.git@branch) fatal: could not read Username for 'https://github.com/': No such device or address
yeah, now the model is loaded, thanks for your support.
use this command its working: pip install git+https://github.com/keras-team/keras-nlp.git
Hi, also I tried installing using [,](pip install git+https://github.com/keras-team/keras-nlp.git), but I still get the same error. is there a workaround or a solution? the error is like: AttributeError: module 'keras_nlp.tokenizers' has no attribute 'compute_word_piece_vocabulary'
Thanks
Hi, tried installing using [,](pip install git+ https://github.com/keras-team/keras-nlp.git), but I still get the same error. when training the tokenizer, is there a workaround or a solution? the error is like: AttributeError: module 'keras_nlp.tokenizers' has no attribute 'compute_word_piece_vocabulary'
regards
On Sat, Aug 27, 2022 at 7:51 AM Naresh Sandrugu @.***> wrote:
use this command its working: pip install git+ https://github.com/keras-team/keras-nlp.git
— Reply to this email directly, view it on GitHub https://github.com/keras-team/keras-io/issues/1031#issuecomment-1229129929, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKDRARVRNRXLCDQ7UJHWBPTV3GUELANCNFSM56UYI4GQ . You are receiving this because you commented.Message ID: @.***>
--
--
DISCLAIMER *** This message and any attachments are confidential and intended solely for the addressee. The following link will display the full disclaimer: http://www.ul.ac.za/disclaimer.jpg http://www.ul.ac.za/disclaimer.jpg ***
--
https://www.ul.ac.za/index.php?Entity=c_events&TheS=970#start https://www.ul.ac.za/index.php?Entity=c_events&TheS=970#start
Thanks, It worked after importing 'compute_word_piece_vocabulary' 👍 [from keras_nlp.tokenizers.word_piece_tokenizer_trainer import ( compute_word_piece_vocabulary,].
But now it complains when executing #StartEndPacker [ TypeError: Exception encountered when calling layer "start_end_packer_1" (type StartEndPacker).
in user code:
File "/usr/local/lib/python3.7/dist-packages/keras_nlp/layers/start_end_packer.py", line 123, in call *
batch_size = tf.shape(inputs)[0]](url)
Please help.
Exactly the same problem I have Naresh, any help, please🤔
On Sat, Aug 27, 2022 at 1:48 PM Naresh Sandrugu @.***> wrote:
am sturck at this point with error any one can help, for reference purpose screen shot below [image: image] https://user-images.githubusercontent.com/59958507/187028860-711c901e-81af-492e-ae4a-9c5f57ac8486.png
— Reply to this email directly, view it on GitHub https://github.com/keras-team/keras-io/issues/1031#issuecomment-1229177581, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKDRARR7VBH53ZWQUEQO3I3V3H6BPANCNFSM56UYI4GQ . You are receiving this because you commented.Message ID: @.***>
--
--
DISCLAIMER *** This message and any attachments are confidential and intended solely for the addressee. The following link will display the full disclaimer: http://www.ul.ac.za/disclaimer.jpg http://www.ul.ac.za/disclaimer.jpg ***
--
https://www.ul.ac.za/index.php?Entity=c_events&TheS=970#start https://www.ul.ac.za/index.php?Entity=c_events&TheS=970#start
yeah, also facing the same issue with start_end_packer function, even i tried with simple examples from keras link like: https://keras.io/api/keras_nlp/layers/start_end_packer/
the below are error,any one can help this
@ygModesto, @Simon-PSR, @NareshSandrugu, linking a comment here, which will solve your issue: https://github.com/keras-team/keras-nlp/issues/301#issuecomment-1221337946. Hope this works! Colab has an older version of TensorFlow pre-installed...so, it has to be uninstalled first.
Thanks Abheesht, start_end_packer issue resolved.
@fchollet we an close this issue