You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
While trying to add dfaligner extract-alignments to the regression test suite, I got a stack dump finishing on KeyError: 'character_tokens' from dfaligner/dataset.py:165.
How to reproduce the bug
cdeveryvoice/tests/regressionsbatchgo.sh# wait until tests finishcdregress-<date>-XXXdfalignertrainconfig/everyvoice-aligner.yaml--config-argstraining.max_steps=1000dfalignerextract-alignmentsconfig/everyvoice-aligner.yaml--modellogs_and_checkpoints/AlignerExperiment/base/checkpoints/last.ckpt
Current environment
Note: the problem was noticed with the following env using torch 2.4.1, but it was reproduced with the same trace with torch 2.3.1, our current torch version.
$ ..everyvoice --diagnostic
EveryVoice Diagnostic information
EveryVoice version: 0.3.0
Python version: 3.10.16 | packaged by conda-forge | (main, Dec 5 2024, 14:16:10) [GCC 13.3.0]
System: Linux 5.15.0-126-generic #136-Ubuntu SMP Wed Nov 6 10:38:22 UTC 2024 x86_64
torch modules installed using pip:
pytorch-lightning==2.5.0.post0
torch==2.4.1
torchaudio==2.4.1
torchinfo==1.8.0
torchmetrics==1.6.1
Bug description
While trying to add
dfaligner extract-alignments
to the regression test suite, I got a stack dump finishing onKeyError: 'character_tokens'
fromdfaligner/dataset.py:165
.How to reproduce the bug
Error messages and logs
Environment
Current environment
Note: the problem was noticed with the following env using torch 2.4.1, but it was reproduced with the same trace with torch 2.3.1, our current torch version.More info
No response
The text was updated successfully, but these errors were encountered: