search

cmusphinx / g2p-seq2seq

G2P with Tensorflow
Other
667 stars 196 forks source link

NotImplementedError: Fast decoding only supports a single shard. #191

Open kannadaraj opened 4 years ago

kannadaraj commented 4 years ago

Hi,

Thanks for this great repo. But this error after trying to use a newly trained model. decoding command g2p-seq2seq --decode 1 --model_dir ./models --return_beams --beam_size 2 Error 

python3.6/site-packages/tensor2tensor-1.7.0-py3.6.egg/tensor2tensor/models/transformer.py", line 497, in _fast_decode
    raise NotImplementedError("Fast decoding only supports a single shard.")
NotImplementedError: Fast decoding only supports a single shard

Please can anyone help?