-
See #19. Targeted data augmentation by cropping images down to specific difficult features can help sculpt GAN outputs by making such images more common, and providing much larger versions which devot…
gwern updated
4 years ago
-
## 🚀 Feature
Standardised interface to pretrain various Transformers with standardised expectations with regards to formatting training data.
## Motivation
To achieve state of the art within…
-
Decription:
Found many bugs during the earlier SotAs BGs..
1-LoS in destroyed gates
2-Passenger players.
3-Passenger player facing.
4-Turret damage scaling
5- Vehicle HP scaling
How it works…
-
There is an existing repo for speech SOTAs: https://github.com/syhw/wer_are_we. Perhaps, you want to reference them and/or join forces with them.
Concerning the Switchboard number, you need to ment…
-
Hi,
I am getting this error while running: python train.py -model DeeperLSTM or python train -model simple_mlp
> img_feature = np.divide(img_feature, np.tile(tem,(1,args.img_vec_dim)))
ValueEr…