2022-12-06 16:44:40,039 (asr_inference_paraformer:337) INFO: decoding, feature length: 412, forward_time: 14.0918, rtf: 3.4203
2022-12-06 16:44:40,077 (asr_inference_paraformer:359) INFO: decoding, predictions: 这样第三产业在快速增长后
2022-12-06 16:44:40,085 (asr_inference_paraformer:324) INFO: decoding, utt_id: ['BAC009S0749W0123']
2022-12-06 16:44:58,486 (beam_search:1300) INFO: decoder input length: 87
2022-12-06 16:44:58,486 (beam_search:1301) INFO: max output length: 11
2022-12-06 16:44:59,051 (beam_search:1380) INFO: adding in the last position in the loop
2022-12-06 16:44:59,053 (beam_search:1316) INFO: no hypothesis. Finish decoding.
2022-12-06 16:44:59,054 (beam_search:1337) INFO: -0.00 * 0.5 = -0.00 for ctc
2022-12-06 16:44:59,054 (beam_search:1340) INFO: total log probability: -1.09
2022-12-06 16:44:59,054 (beam_search:1341) INFO: normalized log probability: -0.08
2022-12-06 16:44:59,054 (beam_search:1342) INFO: total number of ended hypotheses: 10
2022-12-06 16:44:59,054 (beam_search:1344) INFO: best hypo: 目前矿山企业已基本关闭
2022-12-06 16:44:59,055 (asr_inference_paraformer:337) INFO: decoding, feature length: 354, forward_time: 18.9701, rtf: 5.3588
2022-12-06 16:44:59,095 (asr_inference_paraformer:359) INFO: decoding, predictions: 目前矿山企业已基本关闭
2022-12-06 16:44:59,104 (asr_inference_paraformer:324) INFO: decoding, utt_id: ['BAC009S0761W0470']
Traceback (most recent call last):
File "/home/environment/hwang/anaconda3/envs/oslasr/lib/python3.8/runpy.py", line 194, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/home/environment/hwang/anaconda3/envs/oslasr/lib/python3.8/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/home/work_nfs4_ssd/hwang/workspaces/funasr/funasr/bin/asr_inference_launch.py", line 225, in
main()
File "/home/work_nfs4_ssd/hwang/workspaces/funasr/funasr/bin/asr_inference_launch.py", line 219, in main
inference(**kwargs)
File "/home/work_nfs4_ssd/hwang/workspaces/funasr/funasr/bin/asr_inference_paraformer.py", line 329, in inference
results = speech2text(**batch)
File "/home/environment/hwang/anaconda3/envs/oslasr/lib/python3.8/site-packages/torch/autograd/grad_mode.py", line 28, in decorate_context
return func(*args, **kwargs)
File "/home/work_nfs4_ssd/hwang/workspaces/funasr/funasr/bin/asr_inference_paraformer.py", line 193, in call
decoder_outs = self.asr_model.cal_decoder_with_predictor(enc, enc_len, pre_acoustic_embeds, pre_token_length)
File "/home/work_nfs4_ssd/hwang/workspaces/funasr/funasr/models/e2e_asr_paraformer.py", line 333, in cal_decoder_with_predictor
decoder_out, _ = self.decoder(
File "/home/environment/hwang/anaconda3/envs/oslasr/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/home/work_nfs4_ssd/hwang/workspaces/funasr/funasr/models/decoder/transformer_decoder.py", line 504, in forward
x, tgt_mask, memory, memory_mask = decoder(
File "/home/environment/hwang/anaconda3/envs/oslasr/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/home/work_nfs4_ssd/hwang/workspaces/funasr/funasr/models/decoder/transformer_decoder.py", line 123, in forward
x = residual + self.dropout(self.self_attn(tgt_q, tgt, tgt, tgt_q_mask))
File "/home/environment/hwang/anaconda3/envs/oslasr/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/home/work_nfs4_ssd/hwang/workspaces/funasr/funasr/modules/attention.py", line 114, in forward
return self.forward_attention(v, scores, mask)
File "/home/work_nfs4_ssd/hwang/workspaces/funasr/funasr/modules/attention.py", line 83, in forward_attention
scores = scores.masked_fill(mask, min_value)
RuntimeError: The size of tensor a (13) must match the size of tensor b (14) at non-singleton dimension 3
# Accounting: time=2679 threads=1
# Ended (code 1) at Tue Dec 6 16:45:13 CST 2022, elapsed time 2679 seconds
May be, it is a coding bug?