SanghunYun / UDA_pytorch

UDA(Unsupervised Data Augmentation) implemented by pytorch

SanghunYun/UDA_pytorch Issues

apply softmax temperature controlling to augumentation example?
Updated a year ago5
The training stops without triggering errors
Updated 2 years ago10
larger_than_threshold
Updated 2 years ago
Why we have the check global_step > 4999?
Updated 2 years ago
Why unsup_loss is divided by unsup_loss_mask not total?
Updated 3 years ago
Why sup_iter is not repeated when uda_mode is "False"?
Closed 3 years ago1
Does anyone know how total step is set, how much is appropriate, and what is the relationship between total step and epoch?
Updated 3 years ago1
Has anyone run successfully with their own raw dataset??
Updated 3 years ago
how to process new dataset?
Updated 3 years ago1
how port the M to some other clasiffication model like textCNN etc
Updated 3 years ago
TSA(Training Signal Annealing) or Confidence-based masking
Updated 3 years ago
Can the project run on tensorflow 2.0? Because I need to use transformers (it needs to run on tensorflow 2.0)
Updated 3 years ago
Are the quoted accuracies correct in the README?
Closed 3 years ago2
kl divergence uses ori_prob instead of ori_log_prob
Closed 4 years ago1
model.eval() missing?
Updated 4 years ago
Application of dropout when calculating the consistency loss
Updated 4 years ago
Where is the entropy minimization loss ?
Updated 4 years ago
what are the "start" and "end" parameters used when calculating the TSA?
Closed 4 years ago2
Loading Fine-Tuned Pytorch BERT model.
Updated 4 years ago1
calculating the threshold when applying TSA
Closed 4 years ago1
explain how unsup data pre-process works
Updated 5 years ago1
could you provide the example of raw text file format for train, test, unsup?
Closed 5 years ago6