facebookresearch/fairseq

Issues

• DEPRECATION: omegaconf 2.0.6 has a non-standard dependency specifier PyYAML>=5.1.*. pip 24.0 will enforce

  #5436 opened 4 months ago
  1

• FileNotFoundError: [Errno 2] No such file or directory: '/fsx/data/VoxLingua107/manifest/dict.label.txt'

  #5435 opened 5 months ago
  0

• How could we accelerate NLLB, can we use TensorRT or vLLM?

  #5434 opened 5 months ago
  0

• Combining .idx files from fairseq-preprocess

  #5433 opened 5 months ago
  0

• sentence permutation in mBART pretraining

  #5431 opened 5 months ago
  1

• Fair

  #5430 opened 5 months ago
  0

• How to get video features in VideoCLIP without any access to captions

  #5429 opened 5 months ago
  1

• Question for the Facebook research team on XLM applied to new written langauge

  #5424 opened 5 months ago
  0

• WMT'16 En-De Dataset Download Link is broken

  #5423 opened 5 months ago
  3

• How to Train a hubert model with conformer backbone

  #5422 opened 5 months ago
  0

• A clear pynb of all three versions (ASR, TTS, Speech-to-speech) for MMS that can be easily inserted into the Google Colab for converting from audio file to text, text to audio file, and speech-to-speech.

  #5421 opened 5 months ago
  0

• Why the speaking does not pause at period or comma?

  #5420 opened 5 months ago
  0

• Failed to install with pip (Exit code 2)

  #5419 opened 5 months ago
  1

• [Wav2Vec2-XLSR-53]Problems about the pretrained model

  #5418 opened 5 months ago
  1

• Empty and random transcripts after finetuning wav2vec2.

  #5417 opened 5 months ago
  0

• FAILED TO INSTALL

  #5416 opened 5 months ago
  0

• Evaluating the CTC model ,the hypothesises are all the null, resulting in a wer of 100%.

  #5415 opened 5 months ago
  1

• /private/home/changhan/ ... config not found

  #5414 opened 5 months ago
  0

• UST - KMeans Clustering & Acoustic Model(s) Missing

  #5413 opened 5 months ago
  1

• simultaneous agent error? help!!!

  #5412 opened 5 months ago
  0

• convert_wav2vec2 to pytorch

  #5411 opened 5 months ago
  0

• The 'label_smoothed_cross_entropy_with_rdrop.py' file has a bug, preventing validation.

  #5410 opened 5 months ago
  1

• wav2vec2 inference simple process

  #5409 opened 5 months ago
  5

• hi

  #5408 opened 5 months ago
  2

• In load_input_to_shared function of fairseq/fairseq/modules /cuda_utils.cu, does it use a negative index risking with undefined behaviour?

  #5406 opened 6 months ago
  0

• Serious bug in the speech module

  #5405 opened 5 months ago
  1

• Support videoclip with newer version of dependencies

  #5404 opened 6 months ago
  1

• [ASR] Is lexicon-free W2lKenLMDecoder beam search decoding possible with no external language model?

  #5402 opened 6 months ago
  0

• Will the NLLB be upgraded on an ongoing basis？

  #5401 opened 6 months ago
  0

• MMS ASR models without LSAH (language-specific adapters and head) fine-tuning

  #5400 opened 6 months ago
  0

• [Potential Bug] in wav2vec (1.0?): targets always zero, loss computation potentially incorrect?

  #5399 opened 6 months ago
  3

• trainbart.sh error

  #5398 opened 6 months ago
  0

• It seems I am unable to find the error "ValueError: Type mismatch (<class 'int'> vs. <class 'float'>) with values (0 vs. 34.0) for config key: SOLVER.WARMUP_EPOCHS"

  #5397 opened 6 months ago
  0

• Is it okay to directly modify source code ?

  #5396 opened 6 months ago
  0

• Data and instructions for textless_s2st_real_data experiments

  #5395 opened 6 months ago
  0

• How to do W2lKenLMDecoder Word-Level Time Alignment?

  #5394 opened 6 months ago
  1

• ABOUT SRC_TOKENS

  #5393 opened 6 months ago
  0

• SM60 Nvidia P100 is supported or not? Hubert Failing with missing kernel image in forward pass.

  #5392 opened 6 months ago
  0

• Request a new release

  #5391 opened 6 months ago
  0

• how to pretrain x-mod from scratch

  #5390 opened 6 months ago
  2

• ImportError: cannot import name 'hyperparam' from 'sweep' ...

  #5388 opened 7 months ago
  4

• How can I resume training from the dataset point where the model left off

  #5387 opened 7 months ago
  0

• A layer_idx problem in wav2vec2 conformer, need help!!!

  #5386 opened 7 months ago
  3

• Tokenizer token_type_ids all zeros

  #5383 opened 6 months ago
  3

• fairseq-hydra-train couldn't find such file or directory: 'fairseq/data-bin/wikitext-103/dict.txt'

  #5382 opened 7 months ago
  1

• CUDA error of self.embed_positions

  #5381 opened 7 months ago
  1

• Is there any solution to reproduce the data2vec 2.0 pre-trained model with 4 GPUs?

  #5380 opened 7 months ago
  0

• LID model - list of supported languages and source code

  #5379 opened 7 months ago
  0

• About pretraining RoBERTa on custom data

  #5378 opened 7 months ago
  0

• Norwegian support in MMS

  #5377 opened 7 months ago
  0