Open neuwirtter opened 5 months ago
Hello!
I haven't been able to reproduce these issues on my end --- the code currently in the repo is running for me without producing errors.
However, I searched around and was able to find reports of some similar errors (for example, here). Based on this, I'd advise you try either (a) updating the default value for epsilon
in make_ann_classifier
to a real value (1e-7
seems to be the current default value in the docs), or (b) removing the epsilon
and/or clipvalue
arguments from tfk.optimizers.Adam
. Although I cannot reproduce the errors, I can confirm that the script still runs for me with either of these changes.
Hi :),
your solutions worked for me, thank you very much!
Kind regards,
Tereza
čt 16. 5. 2024 v 18:14 odesílatel Akosua @.***> napsal:
Hello!
I haven't been able to reproduce these issues on my end --- the code currently in the repo is running for me without producing errors.
However, I searched around and was able to find reports of some similar errors (for example, here https://github.com/keras-team/keras/issues/9142). Based on this, I'd advise you try either (a) updating the default value for epsilon in make_ann_classifier to a real value (1e-7 seems to be the current default value in the docs https://www.tensorflow.org/api_docs/python/tf/keras/optimizers/Adam), or (b) removing the epsilon and/or clipvalue arguments from tfk.optimizers.Adam. Although I cannot reproduce the errors, I can confirm that the script still runs for me with either of these changes.
— Reply to this email directly, view it on GitHub https://github.com/apbusia/model_based_enrichment/issues/1#issuecomment-2115657225, or unsubscribe https://github.com/notifications/unsubscribe-auth/A4QP4AM6WJG66UW3TQNZ7A3ZCTLPFAVCNFSM6AAAAABHOVZ352VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMJVGY2TOMRSGU . You are receiving this because you authored the thread.Message ID: @.***>
-- Tereza Neuwirthová
Univerzita Karlova v Praze Přírodovědecká fakulta Albertov 6, 128 43 Praha 2 www.natur.cuni.cz
Charles University in Prague Faculty of Science Albertov 6, 128 43 Praha 2 www.natur.cuni.cz/en
Dear developers,
I have tried to use your python script to run tf_example.py model on testing data provided and I've been encountering these error (please see the screeshot, last two errors) - first I solved with modification of the input shape variable (or I thought so: input_shape = (len(seqs.iloc[0]) * len(mbe.AA_ORDER),)) but now I keep getting the second one. Could you please advise on what am I doing wrong?