Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fl_asr_train with ASG criterion fails #1164

Open
harziboy opened this issue Jan 4, 2024 · 0 comments
Open

fl_asr_train with ASG criterion fails #1164

harziboy opened this issue Jan 4, 2024 · 0 comments

Comments

@harziboy
Copy link

harziboy commented Jan 4, 2024

Hi,
when I try to train with ASG criterion, it keeps on terminating with following error:

I0104 05:45:45.500763  2068 Train.cpp:1053] Epoch 1 started!
terminate called after throwing an instance of 'std::invalid_argument'
  what():  FullConnectionCriterion::forward: expects input with dimensions {N, T, B}
*** Aborted at 1704365146 (unix time) try "date -d @1704365146" if you are using GNU date ***
PC: @                0x0 (unknown)
*** SIGABRT (@0x814) received by PID 2068 (TID 0x7f5ca1695000) from PID 2068; stack trace: ***
    @     0x7f5cfa591046 (unknown)
    @     0x7f5ca1042520 (unknown)
    @     0x7f5ca10969fc pthread_kill
    @     0x7f5ca1042476 raise
    @     0x7f5ca10287f3 abort
    @     0x7f5ca14a2b9e (unknown)
    @     0x7f5ca14ae20c (unknown)
    @     0x7f5ca14ae277 std::terminate()
    @     0x7f5ca14ae4d8 __cxa_throw
    @     0x56045beb4154 _ZN2fl3pkg6speech23FullConnectionCriterion7forwardERKNS_8VariableES5_.cold
    @     0x56045bfe478f fl::pkg::speech::AutoSegmentationCriterion::forward()
    @     0x56045bfe4a68 fl::pkg::speech::LinearSegmentationCriterion::forward()
    @     0x56045bfd05e2 _ZZ4mainENKUlSt10shared_ptrIN2fl6ModuleEES_INS0_3pkg6speech17SequenceCriterionEES_INS0_7DatasetEES_INS0_19FirstOrderOptimizerEESA_ddblE5_clES2_S6_S8_SA_SA_ddbl
    @     0x56045bf302ce main
    @     0x7f5ca1029d90 (unknown)
    @     0x7f5ca1029e40 __libc_start_main
    @     0x56045bfbf9c5 _start
    @                0x0 (unknown)

Training data that I am using has been working ok earlier (also with ASG) and does not have issues with CTC criterion.
I have tried to different settings but ASG training always keeps on terminating.
Could there be a general problem with ASG training criterion at the moment?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant