fairseq

fairseq copied to clipboard

Add sequence length padding option for wav2vec2 causes TPU "Scalar type not supported" error during audio pretraining.

1

## 🐛 Bug Add sequence length padding option for wav2vec2 causes TPU "Scalar type not supported" error during audio pretraining. The issue is related to the following commit: https://github.com/pytorch/fairseq/commit/c2b771b1beed56d03134aa0a807fbbec766e484c ....

catalinnega

## ❓ Questions and Help What are the utilities of kaldi_initializer.py and kaldi_decoder.py scripts present inside the fairseq/examples/speech_recognition/kaldi folder?

mukherjeesougata

mBART Continued Pretraining

4

Hello! I am trying to perform continued pretraining on the mbart.cc.25 pretrained checkpoint using the multilingual denoising objective. However, I am not sure how to prepare and pre-process the data...

rohandas14

## ❓ Questions and Help ### Before asking: 1. search the issues. 2. search the docs. #### What is your question? How is the BLEU of the WMT14 test set...

tjshu

## ❓ Questions and Help #### What is your question? How to generate .ctm and .stm files of the decoded texts in wav2vec during speech recognition task? - ubuntu: 20.04...

mukherjeesougata

Hi, In NGramRepeatBlock, the ngram_to_check is defined as ` ngram_to_check = cpu_tokens[bbsz_idx][ -(self.no_repeat_ngram_size - 1) : ] ` The checked tokens must be generated. but there are many padding tokens...

cstsunfu

Is there any method can be used to prune translation model in fairseq?

8

## ❓ Questions and Help ### Before asking: 1. search the issues. 2. search the docs. #### What is your question? #### Code #### What have you tried? #### What's...

robotsp

Is there any way to distill translation models?

4

Prepare to compress the translation model trained by fairseq, such as distillation and pruning. Can you give me some advice? Thank U.

AIikai

Unable to do inference

4

I am unable to do inference. I am running the command : python infer.py dataset_dec_dev-other/ --task audio_pretraining --nbest 1 --path wav2vec_vox_960h_pl.pt --gen-subset train --results-path results_decode_dev-other --w2l-decoder viterbi --sil-weight 0 --criterion...

pratikkumar018

Wav2vec2 error after validation when training : terminate called after throwing an instance of 'c10::Error'

14

## 🐛 Bug Wav2vec2 error after validation when training ### To Reproduce Steps to reproduce the behavior (**always include the command you ran**): TASK=audio_pretraining CRITERION=ctc MAX_TOKENS=960000 #about 16.6 sentences #Optimization...

Jxu-Thu