facebookresearch/fairseq2

Issues

• Compile a matmul + reduction op in vector quantizer to avoid excess memory usage.

  #832 opened 2 months ago by shagunsodhani
  2

• Add Sophia optimizer

  #852 opened 9 days ago by PeanutButterRat
  0

• Add GSM8K recipe examples

  #848 opened 15 days ago by zyaoj
  0

• Build Failure When Running cmake --build build

  #842 opened 23 days ago by shashi-23
  0

• Support checksum in AssetDownloadManager

  #839 opened a month ago by cbalioglu
  0

• Cannot install fairseq2 on WSL (versions have conflicting dependencies)

  #786 opened 2 months ago by dmehta-huro
  1

• documentation and training recipes

  #682 opened 4 months ago by zhhao1
  1

• Dockerfile and/or Docker image

  #783 opened 2 months ago by alvations
  0

• pip3 install fairseq2 - doesn't work

  #471 opened 7 months ago by twerff
  6

• Which commit corresponds to the PyPi 0.2.1 release?

  #769 opened 3 months ago by boldorider4
  0

• Fairseq 0.2.1 - No wheels available for PyTorch 2.3.* and 2.4.*

  #745 opened 3 months ago by julianzaidi
  1

• Introduce CLI Interface `eval` for Model Evaluation

  #689 opened 4 months ago by Ahmedsaed
  1

• Create fairseq2 LM Evaluation Harness Wrapper

  #707 opened 4 months ago by mostafaelhoushi
  0

• add without_repeatition option to DataPipeline.round_robin and DataPipeline.sample methods

  #685 opened 4 months ago by artemru
  0

• load an Audio with fairseq2.data.AudioDecoder with specific offset and duration

  #661 opened 4 months ago by alihejazi97
  0

• [LayerSkip] Self-Speculative Decoding

  #642 opened 4 months ago by mostafaelhoushi
  0

• [LayerSkip] Early Exit Loss

  #641 opened 4 months ago by mostafaelhoushi
  0

• [LayerSkip] Per-Layer Dropout Rate Configuration

  #640 opened 4 months ago by mostafaelhoushi
  0

• CLI crashes when executed without arguments

  #629 opened 5 months ago by Ahmedsaed
  0

• Dynamic bucketing

  #453 opened 5 months ago by artemru
  0

• Unexpected behavior when using repeat() with sample()/round_robin()

  #604 opened 5 months ago by syleshfb
  0

• fairseq2n is incompatible with torch 2.3.0

  #600 opened 5 months ago by 0de554K
  0

• AMD compilation error

  #583 opened 5 months ago by taikai-zz
  2

• When I compiled the fairseq2n native source code, an error occurred

  #569 opened 6 months ago by skywindy
  0

• Write w2v-BERT pretraining recipe

  #313 opened 9 months ago by cbalioglu
  2

• Introduce model parallelism

  #316 opened 9 months ago by cbalioglu
  1

• `ModelLoader` doesn't allow loading a local checkpoint

  #456 opened 7 months ago by zrthxn
  1

• DataPipelineBuilder.shard should have an argument to keep the remainder

  #351 opened 7 months ago by artemru
  1

• seed argument in DataPipelineBuilder.shuffle

  #350 opened 7 months ago by artemru
  1

• BuilderDataPipeline.repeat

  #352 opened 7 months ago by artemru
  1

• pre-built packages for PyTorch 2.2.0 and 2.1.2 do not exist on FAIR's package repository

  #356 opened 9 months ago by Alrowithi
  1

• Does fairseq2 support command tools like fairseq?

  #383 opened 7 months ago by piwawa
  1

• where is checkpoint directory?

  #427 opened 8 months ago by aknvictor
  1

• DatapipelineBuilder.map should not implicitly convert python object

  #375 opened 8 months ago by artemru
  2

• Raw source for fairseq2n code

  #370 opened 9 months ago by natgillin
  4

• using pipeline_builder shared pointer multiple times lead to segfaults

  #369 opened 9 months ago by artemru
  1

• DataPipeline execution profiling

  #355 opened 9 months ago by artemru
  0

• operation priority in DataPipeline operations

  #354 opened 9 months ago by artemru
  0

• `from_generator` function for BuilderDataPipeline

  #353 opened 9 months ago by artemru
  0

• ImportError: cannot import name 'SequenceToTextOutput' from 'fairseq2.generation'

  #349 opened 9 months ago by gangbanyegeirichuan
  0

• S2UT model pre-training using pretrained wav2vec 2 showing dimension mismatch

  #347 opened 9 months ago by mahendraphd
  0

• Using WaveformToFbankConverter with variable sample rates is impossible

  #341 opened 9 months ago by avidale
  0

• Unable to run fairseq2 inside nvidia/cuda docker container due to libfairseq2n.so.0: undefined symbol error

  #321 opened 9 months ago by kyurkchyan
  0

• Introduce support for StreamingLLM

  #320 opened 9 months ago by cbalioglu
  0

• Start using iwyu in C++ code base

  #319 opened 9 months ago by cbalioglu
  0

• Add logging to C++ code base

  #318 opened 9 months ago by cbalioglu
  0

• Run CI tests on GPU

  #314 opened 9 months ago by cbalioglu
  0

• Introduce sequence parallelism

  #317 opened 9 months ago by cbalioglu
  0

• Downstream w2v-BERT improvements from SC

  #312 opened 9 months ago by cbalioglu
  0

• Use fused RMSNorm

  #315 opened 9 months ago by cbalioglu
  0