issues
search
ybracke
/
transnormer
A lexical normalizer for historical spelling variants using a transformer architecture.
GNU General Public License v3.0
6
stars
1
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Dev change save and logging strategy
#97
ybracke
closed
1 month ago
0
Dev extend evaluation
#96
ybracke
closed
1 month ago
0
Dev fix evaluation
#95
ybracke
closed
1 month ago
0
mT5 Tokenizer
#94
ybracke
opened
4 months ago
0
Evaluation: Code for analysis
#93
ybracke
opened
5 months ago
0
Evaluation: Allow ignoring types
#92
ybracke
opened
7 months ago
0
WIP: Draft for Trainer with custom logging
#91
ybracke
opened
7 months ago
0
Do we need `[beam_search_decoding]` in training configs?
#90
ybracke
opened
7 months ago
0
Improve configuration of training arguments
#89
ybracke
opened
7 months ago
0
Turn limitation of memory usage into a configurable property
#88
ybracke
opened
7 months ago
0
Support to choose running training/generation on CPU
#87
ybracke
opened
7 months ago
0
Remove or update `split_dataset`
#86
ybracke
opened
8 months ago
0
Evaluation: Move `add_sent_scores.py` functionality into `evaluate.py`
#85
ybracke
opened
8 months ago
0
Dev evaluation
#84
ybracke
closed
8 months ago
0
Evaluation: Output example specific scores
#83
ybracke
closed
8 months ago
0
Write tests for evaluation functions
#82
ybracke
opened
8 months ago
0
Refactor data processing in train_model
#81
ybracke
closed
8 months ago
0
Improve data handling for training and inference
#80
ybracke
closed
8 months ago
0
Estimate the difficulty of each dataset
#79
ybracke
opened
8 months ago
0
Exclude long input sequences during training
#78
ybracke
closed
8 months ago
1
Make ROOT variable no longer hard-coded
#77
ybracke
closed
8 months ago
0
Efficient generation
#76
ybracke
closed
7 months ago
0
Use a GenerationConfig for generating normalizations
#75
ybracke
closed
8 months ago
0
Use a `GenerationConfig` for generation
#74
ybracke
closed
8 months ago
0
Integrate into MONAPipe
#73
ybracke
opened
8 months ago
0
Dev evaluate
#72
ybracke
closed
10 months ago
0
Evaluation: Trace metrics/predictions to train and test parameters
#71
ybracke
opened
10 months ago
1
Dev prediction script
#70
ybracke
closed
10 months ago
0
Restructuring/extending README and doc
#69
ybracke
closed
10 months ago
0
Support for byT5 models (tokenizer-free)
#68
ybracke
closed
11 months ago
0
Improve implementation for different model types
#67
ybracke
opened
11 months ago
1
Implement pre-commit hooks
#66
ybracke
closed
11 months ago
0
Efficient training
#65
ybracke
opened
1 year ago
4
Hyphens
#64
ybracke
opened
1 year ago
2
Experiment with smaller versions of pre-trained models
#63
ybracke
opened
1 year ago
0
Improve training data: post-correct CAB-normalized DTA texts
#62
ybracke
closed
9 months ago
2
Evaluation: Implement evaluation
#61
ybracke
closed
10 months ago
1
Add a template for inspect predictions
#60
ybracke
closed
1 year ago
0
Set up using the DVC registry of the ZDL as remote storage
#58
ybracke
opened
1 year ago
0
Experiment with randomly initialized encoder
#55
ybracke
opened
1 year ago
1
Implement custom `tokenizers.decoders.Decoder` for output serialization
#53
ybracke
opened
1 year ago
0
`main` function for `train_model.py`
#52
ybracke
opened
1 year ago
0
Evaluation: Get loss
#51
ybracke
opened
1 year ago
0
Language cleaning
#50
ybracke
opened
1 year ago
1
Dealing with long input sequences
#49
ybracke
opened
1 year ago
2
Loading functionality for Anselm corpus
#48
ybracke
opened
1 year ago
0
Additional gold training data
#47
ybracke
opened
1 year ago
15
Refactor train_model by moving steps into functions
#46
ybracke
closed
1 year ago
0
Add the option to use a randomly initialized encoder/decoder instead of a pretrained model
#45
ybracke
opened
1 year ago
0
Synthetic training data
#44
ybracke
opened
1 year ago
1
Next