fairseq2
fairseq2 copied to clipboard
facebookresearch
FAIR Sequence Modeling Toolkit 2
**What does this PR do? Please describe:** Adds support for wav2vec2 large models trained on Librispeech 960h (ls960) and LibriVox 60k (lv60k). Tested with this simple code snippet: lv60k ```...
**What does this PR do? Please describe:** A summary of the change or the issue that is fixed. Fixes #{issue number} **Does your PR introduce any breaking changes? If yes,...
Is there any documentation or `examples` that I can refer to train a transformer model from scratch using `fairseq2`? The `examples` folder in the repository seems empty.
As the second recipe after NLLB, write the w2v-BERT (and wav2vec2) pretraining recipe for users to check out. This will likely branch to several subtasks once we start working on...
Introduce Megatron style model parallelism.
**What does this PR do? Please describe:** I added python 3.12 support to prebuilt wheels. Unable to test locally at the moment. Fixes #471 **Does your PR introduce any breaking...
I'm trying to install seamlessm4q on my Macbook M2. Installation results in an error related to fairseq2. So, I tried to manually install fairseq using `% pip3 install fairseq2`. It...
**Is your feature request related to a problem? Please describe:** Samples participating in data pipeline can possess different characteristics (like length, number of tokens, ...). Currently we can only bucket...
I tried to install the pre-built package from FAIR's package repo. But apparently it's missing for PyTorch 2.2.0 and PyTorch 2.1.2 (although in the combination matrix it exists) and so...
**What does this PR do? Please describe:** Implements `video_decoder` for loading a video dataset as part of the data_pipeline using ffmpeg libraries. The libraries used are libavcodec, libavformat, libavutil, and...
