wwfcnu
wwfcnu
Is G too big?How big G can it support
> Can you using https://github.com/k2-fsa/icefall/blob/master/egs/librispeech/ASR/local/compile_hlg_using_openfst.py > > Your G is too large and I am afraid it won't fit into your GPU memory even if you can build an HLG...
> > Can you using https://github.com/k2-fsa/icefall/blob/master/egs/librispeech/ASR/local/compile_hlg_using_openfst.py > > Your G is too large and I am afraid it won't fit into your GPU memory even if you can build an...
/usr/local/python3.7.10/lib/python3.7/site-packages/torchaudio/_internal/module_utils.py:99: UserWarning: Failed to import soundfile. 'soundfile' backend is not available. warnings.warn("Failed to import soundfile. 'soundfile' backend is not available.") tools/k2/make_hlg.sh: line 26: 1258 Segmentation fault (core dumped) python tools/k2/prepare_char.py...
> > > tools/k2/make_hlg.sh: line 38: 1331 Segmentation fault (core dumped) python > > > > > > Could you run > > ``` > > export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python > >...
> You can ignore that warning. ok
> You can either use `compile_hlg_using_openfst.py` (if your G is very large and if `compile_hlg.py` throws OOM) or `compile_hlg.py`. > > You can choose any of them if they both...
I used make_hlg.sh before
> > I used make_hlg.sh before > > I think `make_hlg.sh` invokes `compile_hlg.py`. You can replace it with `compile_hlg_using_openfst.py` if you encounter OOM due to using a large G. but...
> ok,thanks