issues
search
sanchit-gandhi
/
seq2seq-speech
Repository for fine-tuning Transformers 🤗 based seq2seq speech models in JAX/Flax.
34
stars
6
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
[Eval] Compute WER and CER, perform eval at end of epoch if eval steps is unspecified
#48
sanchit-gandhi
closed
2 years ago
0
[CTC Tokenizer] Gigaspeech, SWB and AMI
#47
sanchit-gandhi
closed
2 years ago
0
[Eval] Update logger to log wer or cer (depending on data args)
#46
sanchit-gandhi
closed
2 years ago
0
[Eval] Evaluate with wer (default) or cer metrics
#45
sanchit-gandhi
closed
2 years ago
0
Remove spaced apostrophes and <unk> token
#44
sanchit-gandhi
closed
2 years ago
0
[Eval] Change default split name + set steps to max_steps
#43
sanchit-gandhi
closed
2 years ago
0
make style
#42
sanchit-gandhi
closed
2 years ago
0
[Train] Fix bug in reference to epochs
#41
sanchit-gandhi
closed
2 years ago
0
[CTC] Add beam search decoder
#40
sanchit-gandhi
closed
2 years ago
0
[Train] Remove speech disfluencies and clean data
#39
sanchit-gandhi
closed
2 years ago
0
[TEDLIUM] Remove speech disfluencies
#38
sanchit-gandhi
closed
2 years ago
0
[CTC] Remove unnecessary arg `ignore_mismatched_size`
#37
sanchit-gandhi
closed
2 years ago
0
[CTC] Fix bug in model-tokenizer vocab size mismatch
#36
sanchit-gandhi
closed
2 years ago
0
[CTC] Remove non-italics apostrophes and speech marks from train set
#35
sanchit-gandhi
closed
2 years ago
0
Train for N steps, eval every M and save every K
#34
sanchit-gandhi
closed
2 years ago
0
[CTC] Remove punctuation from train split only
#33
sanchit-gandhi
closed
2 years ago
0
[CTC] Facilitate removal of punctuation from tokenizer
#32
sanchit-gandhi
closed
2 years ago
0
Fix bug in data arg
#31
sanchit-gandhi
closed
2 years ago
0
[CTC] Enable removal of punctuation
#30
sanchit-gandhi
closed
2 years ago
0
Update CTC notebook
#29
sanchit-gandhi
closed
2 years ago
0
Negative Losses in CTC Training
#28
sanchit-gandhi
closed
2 years ago
2
Check Negative CTC Loss
#27
sanchit-gandhi
closed
2 years ago
0
Update run_ctc_dummy.sh
#26
sanchit-gandhi
closed
2 years ago
0
Make CTC tokenizer `do_lower_case` attribute water-tight
#25
sanchit-gandhi
closed
2 years ago
0
Fix `do_lower_case` bug in CTC train script
#24
sanchit-gandhi
closed
2 years ago
0
CTC tokenizer returns <unk> tokens with `do_lower_case=True`
#23
sanchit-gandhi
closed
2 years ago
1
Add `--do_predict` to CTC script
#22
sanchit-gandhi
closed
2 years ago
0
Fix`train_metrics` bug in CTC script
#21
sanchit-gandhi
closed
2 years ago
0
Pad target to max length if int is not specified
#20
sanchit-gandhi
closed
2 years ago
0
Run train/eval/test in one-script
#19
sanchit-gandhi
closed
2 years ago
0
File to create CTC Tokenizer
#18
patrickvonplaten
closed
2 years ago
0
Wav2Vec2 CTC in Flax/JAX
#17
patrickvonplaten
closed
2 years ago
0
Training script clean-up
#16
sanchit-gandhi
closed
2 years ago
0
Beam search with scan
#15
sanchit-gandhi
closed
2 years ago
0
Fuse matmul operations
#14
sanchit-gandhi
closed
2 years ago
0
Gradient checkpointing and scan
#13
sanchit-gandhi
closed
2 years ago
0
Correct Mixed Precision Training
#12
sanchit-gandhi
closed
2 years ago
0
[train] Optimise gradient accumulation loop
#11
sanchit-gandhi
closed
2 years ago
0
[feat] Save and use feature encoder outputs during training
#10
sanchit-gandhi
closed
2 years ago
0
[FlaxWav2Vec2Model] Fix bug in attention mask
#9
sanchit-gandhi
closed
2 years ago
0
[FlaxSpeechEncoderDecoder] Fix input shape bug in weights init
#8
sanchit-gandhi
closed
2 years ago
0
(feat) Clip Gradients by Global Norm
#7
sanchit-gandhi
closed
2 years ago
0
Update README.md
#6
sanchit-gandhi
closed
2 years ago
0
Convert Flax training script to a standalone training script
#5
sanchit-gandhi
closed
2 years ago
0
Convert Flax SpeechEncoderDecoderModel to a standalone model script
#4
sanchit-gandhi
closed
2 years ago
0
Convert Flax Wav2Vec2 to a standalone model script
#3
sanchit-gandhi
closed
2 years ago
0
Convert Flax Bart to a standalone model script
#2
sanchit-gandhi
closed
2 years ago
0
Add modelling files from Transformers 🤗
#1
sanchit-gandhi
closed
2 years ago
0
Previous