Top transformers issues

Proposal: seq2seq tokenizers expose a prepare_seq2seq_batch method

Tokenization
sshleifer picture sshleifer·28 Jul 2020·11 Comments

Fine tune T5 for paraphrase generation

wontfix
mengyahuUSTC-PU picture mengyahuUSTC-PU·24 Jul 2020·7 Comments

google/reformer-enwik8 tokenizer was not found in tokenizers model name list

wontfix
pzelasko picture pzelasko·14 Jul 2020·4 Comments
👍2

Add beta 1 and beta 2 option in `TrainingArguments` for `AdamW` optimizer.

PhilipMay picture PhilipMay·12 Jul 2020·3 Comments

Easier way to download pretrained model files to local

wontfix
m0hit-aggarwal picture m0hit-aggarwal·6 Jul 2020·1 Comments
👍4

Get prediction_scores from BART forward method

mmsamiei picture mmsamiei·4 Jul 2020·3 Comments
👍2

Modify BERT/BERT-descendants to be TorchScript-able (not just traceable)

sbrody18 picture sbrody18·16 Jun 2020·16 Comments
👍2

🐛 TFTrainer not working on TPU (TF2.2)

TensorFlow wontfix
astariul-colanim picture astariul-colanim·15 Jun 2020·3 Comments

Add Linformer model

New model wontfix
AaronFriel picture AaronFriel·13 Jun 2020·4 Comments
🚀22 🎉15 👀11

[generate] Increasing length_penalty makes generations longer

Discussion seq2seq wontfix
sshleifer picture sshleifer·10 Jun 2020·4 Comments
 Previous issues
Next issues