patrickvonplaten
Follow
Create your own GitHub profile
Sign up for your own profile on GitHub, the best place to host code, manage projects, and build software alongside 50 million developers.
Sign upPopular repositories
1,585 contributions in the last year
Contribution activity
November 2020
Created 3 repositories
- patrickvonplaten/rustlings Rust
- patrickvonplaten/google-research Jupyter Notebook
-
patrickvonplaten/hugging_face_challenge
Jupyter Notebook
•
Built by
Created a pull request in huggingface/tokenizers that received 7 comments
Split Pre-Tokenizer
Fix #468
This is very much work in progress and a first version on how a Split pre-tokenizer can be implemented.
I'm still very inexperienced in Ru…
+494
−2
•
7
comments
Opened 27 other pull requests in 4 repositories
huggingface/transformers
20
merged
2
open
1
closed
- [MT5] Add use_cache to config
- [Flax test] Add require pytorch to flix flax test
- Fix dpr<>bart config for RAG
- Revert "[s2s] finetune.py: specifying generation min_length"
- [Pegasus] Refactor Tokenizer
- [Generate Test] fix flaky ci
- [Tokenizer Doc] Improve tokenizer summary
- [MT5] More docs
- [XLNet] Fix mems behavior
- T5 & mT5
- [T5] Fix load weights function
- [T5] Bug correction & Refactor
- [WIP] T5v1.1 & MT5
- [T5Tokenizer] fix t5 token type ids
- [T5 Tokenizer] Fix t5 special tokens
- [Tests] Add Common Test for Training + Fix a couple of bugs
- [TF generate] Cut encoder outptus to just last hidden states for now
- [WIP] Add performer in flax
- [Generate Test] fix greedy generate test
- [Seq2Seq] Make Seq2SeqArguments an independent file
- [Seq2Seq] Correct import in Seq2Seq Trainer
- [Notebooks] Add new encoder-decoder notebooks
- [EncoderDecoder] fix encoder decoder config model type bug
huggingface/datasets
1
open
1
merged
google-research/google-research
1
open
huggingface/blog
1
merged
Reviewed 64 pull requests in 4 repositories
huggingface/transformers 59 pull requests
- Attempt to fix Flax CI error(s)
- Better filtering of the model outputs in Trainer
- Add T5 Encoder for Feature Extraction
- [Pegasus] Refactor Tokenizer
- Fix embeddings resizing in TF models
- Create README.md
- BART & FSMT: fix decoder not returning hidden states from the last layer
- Fix dpr<>bart config for RAG
- MPNet: Masked and Permuted Pre-training for Language Understanding
- [s2s finetune trainer] potpurri of small fixes
- Fix PPLM
- [FlaxBert] Fix non-broadcastable attention mask for batched forward-passes
- [XLNet] Fix mems behavior
- Better booleans handling in the TF models
- [s2s] finetune.py: specifying generation min_length
- Return correct Bart hidden state tensors
- fix rag index names in eval_rag.py example
- MT5 should have an autotokenizer
- [core] implement support for run-time dependency version checking
- Rewrite ProphetNet to adapt converting ONNX friendly
- Fix bug in x-attentions output for roberta and harden test to catch it
- Add FastFormers to the example directory
- Support various BERT relative position embeddings (2nd)
- Vectorize RepetitionPenaltyLogitsProcessor to improve performance
- moved temperature warper before topP/topK warpers
- Some pull request reviews not shown.
huggingface/datasets 3 pull requests
huggingface/tokenizers 1 pull request
huggingface/widgets-server 1 pull request
Created an issue in huggingface/tokenizers that received 2 comments
Link to docs in "About" section of repo
Can we add a link to https://huggingface.co/docs/tokenizers/python/latest/ in the About section similar to how it's done for Transformers:
2
comments