- 08 Apr, 2021 1 commit
-
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 367463455
-
- 07 Apr, 2021 1 commit
-
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 367244904
-
- 06 Apr, 2021 7 commits
-
-
Reed Wanderman-Milne authored
PiperOrigin-RevId: 367105004
-
Hongkun Yu authored
PiperOrigin-RevId: 367101911
-
Xianzhi Du authored
PiperOrigin-RevId: 367083514
-
Reed Wanderman-Milne authored
This has no functional impact since the default is currently True, but I plan on changing the default to False soon. PiperOrigin-RevId: 367049942
-
Reed Wanderman-Milne authored
The function `tf.train.experimental.enable_mixed_precision_graph_rewrite` will be removed from the TF2 namespace soon, at which point it will only be accessible under tf.compat.v1. PiperOrigin-RevId: 367046393
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 366939051
-
Jeremiah Liu authored
For the `GaussianProcessClassificationHead`, the temperature scaling needs to be disabled during training to avoid unexpected modification to the learning rate, which harms model quality. (Unfortunately, this seems to require adding `training` to the `call` method). Also set the default of `gp_cov_ridge_penalty` in `RandomFeatureGaussianProcess` to 1 to be consistent with that in the `GaussianProcessClassificationHead`. PiperOrigin-RevId: 366917075
-
- 05 Apr, 2021 12 commits
-
-
Scott Zhu authored
PiperOrigin-RevId: 366900579
-
Reed Wanderman-Milne authored
This shouldn't break any official models, since I changed all LossScaleOptimizer isinstance checks to use the nonexperimental version (the experimental LSO subclasses the nonexperimental LSO, so changing isinstance checks in this way is always safe). PiperOrigin-RevId: 366891847
-
Jeremiah Liu authored
This change allows `GaussianProcessClassificationHead` to output only the predictive logits during training and evaluation (instead of outputting a tuple `(logits, covmat)`). The goal is to make the layer more compatible with `SentencePredictionTask` and Keras' `model.fit()` API. PiperOrigin-RevId: 366891298
-
Fan Yang authored
PiperOrigin-RevId: 366889393
-
Allen Wang authored
PiperOrigin-RevId: 366883662
-
Hongkun Yu authored
PiperOrigin-RevId: 366883391
-
Hongkun Yu authored
PiperOrigin-RevId: 366883351
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 366879385
-
A. Unique TensorFlower authored
Rewrite the GIOU loss op in matrix form to avoid the map_fn. map_fn could be prohibitively costly when dealing with large number of bounding boxes. PiperOrigin-RevId: 366854232
-
Vighnesh Birodkar authored
PiperOrigin-RevId: 366850188
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 366819679
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 366817283
-
- 04 Apr, 2021 1 commit
-
-
Pablo Ribalta Lorenzo authored
Signed-off-by:Pablo Ribalta Lorenzo <pribalta@nvidia.com>
-
- 03 Apr, 2021 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 366540340
-
- 02 Apr, 2021 9 commits
-
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 366496850
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 366395416
-
Vighnesh Birodkar authored
PiperOrigin-RevId: 366395416
-
Abdullah Rashwan authored
PiperOrigin-RevId: 366391425
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 366382220
-
Vighnesh Birodkar authored
PiperOrigin-RevId: 366388829
-
Vighnesh Birodkar authored
PiperOrigin-RevId: 366385685
-
Vighnesh Birodkar authored
PiperOrigin-RevId: 366382220
-
Frederick Liu authored
PiperOrigin-RevId: 366368802
-
- 01 Apr, 2021 6 commits
-
-
Chen Chen authored
PiperOrigin-RevId: 366354480
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 366330392
-
Chen Chen authored
PiperOrigin-RevId: 366313554
-
Jeremiah Liu authored
PiperOrigin-RevId: 366178203
-
Jeremiah Liu authored
PiperOrigin-RevId: 366175727
-
Jeremiah Liu authored
PiperOrigin-RevId: 366170373
-
- 31 Mar, 2021 2 commits
-
-
Ronny Votel authored
PiperOrigin-RevId: 366134194
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 366120838
-