patrickvonplaten
Follow
patrickvonplaten
Follow
Popular repositories
-
Wav2Vec2_PyCTCDecode Public
Small repo describing how to use Hugging Face's Wav2Vec2 with PyCTCDecode
-
-
-
-
1,519 contributions in the last year
Less
More
Contribution activity
February 2022
Created 22 commits in 4 repositories
Created 1 repository
Created a pull request in huggingface/transformers that received 44 comments
TF generate refactor - Greedy Search
What does this PR do?
This PR is the first step to refactor TF's generate() method similar in spirit to https://discuss.huggingface.co/t/big-genera…
+1,475
−206
•
44
comments
Opened 11 other pull requests in 1 repository
huggingface/transformers
1
open
10
merged
- [Discuss][Wav2Vec2 Time Stamps]
- [Wav2Vec2ProcessorWithLM] Fix auto processor with lm
- Add section about doc testing
- [SpeechEncoderDecoder] Make sure no EOS is generated in test
- Small clean up generate
- Add example batch size to all commands
- [ASR pipeline] correct asr pipeline for seq2seq models
- Revert "Handle PyTorch to Flax conversion of 1D convolutions"
- [Flax tests] Disable scheduled GPU tests
- [torch_int_div] Correct true division in generation
- [Wav2Vec2ProcessorWithLM] add alpha & beta to batch decode & decode
Reviewed 45 pull requests in 4 repositories
huggingface/transformers
41 pull requests
- Fix shape
- Add PLBart
- [benchmark tool] trainer-benchmark.py
- [GPT2] Add lm_attention_mask as a optional argument
- Implementation of activations as pytorch modules
- TF generate refactor - Greedy Search
- Add section about doc testing
- Updated the RAG training with latest Pytorch Lightning library and the RAY
-
Add
decoder_kwargsto send to LM on asr pipeline. - Update bad_words_ids usage
- Fix typo in speech2text2 doc
- Fix ASR pipelines from local directories with wav2vec models that have language models attached
- [WIP] Flax Speech-Encoder-Decoder Model
- Fix Seq2SeqTrainer for VisionEncoderDecoderModel
- Fix bugs of s2t fairseq model converting
- TF MT5 embeddings resize
- update serving_output for some TF models
- Add implementation of typical sampling
- [Flax tests/FlaxBert] make from_pretrained test faster
- Add Data2Vec
- Upgrade black to version ~=22.0
- Add Wav2Vec2 Adapter Weights to Flax
- [Flax tests] fix test_model_outputs_equivalence
- PoC for a ProcessorMixin class
- feat(flax): allow encoder_outputs in generate
- Some pull request reviews not shown.
huggingface/datasets
2 pull requests
huggingface/blog
1 pull request
huggingface/huggingface_hub
1 pull request
Created an issue in google-research/google-research that received 8 comments
[Scaling Efficiently: Insights from Pre-training and Finetuning Transformers] Some checkpoints seem to be incomplete or incorrectly uploaded
Hey @vanzytay, Thanks a lot for uploading all of your newest T5 checkpoints to GCP. It's really amazing that you guys are taking the time to make t…
8
comments
