Hello, after I downloaded the 5-gram Language model, I ran the corresponding eval_competition file and configured the lm_decoder parameter as follows:
nbest = lmDecoderUtils.lm_decode(
ngramDecoder,
logits[0],
blankPenalty=blank_penalty,
returnNBest=True,
rescore=True,
)
it will show the fault "Segmentation fault (core dumped)". However it will operate successfully when rescore=False. May I ask how this situation should be resolved. Thank you!