Aborted error when attempting to use ctc-decoding about icefall HOT 2 CLOSED

I receive an error trying to decode using ctc-decoding

ctc-decoding is tested with a model using vocab size 500. To fix

[F] /usr/share/miniconda/envs/k2/conda-bld/k2_1631619831677/work/k2/csrc/array.h:501:void k2::Array1<T>::Init(k2::ContextPtr, int32_t, k2::Dtype) [with T = char; k2::ContextPtr = std::shared_ptr<k2::Context>; int32_t = int] Check failed: size >= 0 (-1383015021 vs. 0) Array size MUST be greater than or equal to 0, given :-1383015021

please change

 modified=True, 

if you are using a model with vocab size 5000.

Also, you may need to reduce the value of --max-duration.
(Currently, you are using 300. It is OK for a model with vocab size 500).

Sorry that this is not documented. There is an ongoing PR to fix it (https://github.com/k2-fsa/icefall/pull/91/files)
and there are discussions about it (#70 (comment))

As for the second error,

_pickle.UnpicklingError: invalid load key, 'v'

I think you skipped an essential step listed in https://icefall.readthedocs.io/en/latest/recipes/librispeech/conformer_ctc.html#download-the-pre-trained-model

git lfs install

To fix that, please first install git-lfs (using sudo apt-get install git-lfs if you are using Ubuntu),
go to the pre-trained model repo and run

cd icefall_asr_librispeech_conformer_ctc
git lfs pull

After that, everything should work.

BTW, please use the pre-trained model in https://github.com/csukuangfj/icefall-asr-conformer-ctc-bpe-500
to test ctc-decoding if you don't modify any code.

from icefall.

Thank you this worked!

from icefall.