Commits

History for

Commits on Mar 6, 2017

Add NoGradients for QuantizeAndDequantizeV2Op's input_min and input_max arguments.

Suharsh Sivakumar
authored and
tensorflower-gardener
committed

Commits on Feb 18, 2017

Merge changes from github.

Vijay Vasudevan
authored and
tensorflower-gardener
committed

Commits on Feb 15, 2017

Make input_min and input_max Inputs of QuantizeAndDequantize rather than attributes.

Suharsh Sivakumar
authored and
tensorflower-gardener
committed

Commits on Feb 7, 2017

Better error message from trying to differentiate PreventGradient.

alextp
authored and
tensorflower-gardener
committed

Commits on Jan 11, 2017

Switch tf.concat_v2 references in third_party/tensorflow to tf.concat.
tensorflower-gardener
committed

Commits on Jan 6, 2017

Add support for gradient calculation for concat_v2 with negative axis to fix github
tensorflower-gardener
committed

Commits on Jan 4, 2017

Adds the following new ops:
tensorflower-gardener
committed

Commits on Dec 22, 2016

Allows int64 indices in gather_nd gradient.
tensorflower-gardener
committed

Commits on Dec 16, 2016

Replace array_ops.unpack with array_ops.unstack in third_party/tensorflow.
tensorflower-gardener
committed

Commits on Dec 15, 2016

Switch array_ops.pack/unpack to array_ops.stack/unstack. Also switch a few remaining references to tf.pack/unpack to tf.stack/unstack.
tensorflower-gardener
committed

Commits on Dec 14, 2016

Switch leftover reference to tf.concat to tf.concat_v2 in array_grad.py.
tensorflower-gardener
committed

Commits on Dec 13, 2016

Change implementation of concat grad to use new split. Results in up to a 5x
tensorflower-gardener
committed

Commits on Dec 9, 2016

Commits on Dec 7, 2016

Remove remaining uses of tf.reverse() and remove tf.reverse()

aselle
authored and
tensorflower-gardener
committed

Commits on Dec 1, 2016

Rename more argument names to be consistent with NumPy.

aselle
authored and
tensorflower-gardener
committed

Commits on Nov 22, 2016

Add concat_v2 to array_ops.py. Also, add gradient support for concat_v2 op.
tensorflower-gardener
committed

Commits on Nov 17, 2016

Merge changes from github.
tensorflower-gardener
committed

Commits on Nov 15, 2016

Expose tf.reverse_v2 which takes indices rather than a bool tensor.

aselle
authored and
tensorflower-gardener
committed

Commits on Nov 11, 2016

Add a new op split_v that can handle variable size splits.
tensorflower-gardener
committed

Commits on Nov 3, 2016

Adding CPU kernels for tf.scatter_nd(), tf.scatter_nd_update(), tf.scatter_nd_add(), tf.scatter_nd_sub(), tf.scatter_nd_mul() and tf.scatter_nd_div() as well as gradient functions for tf.scatter_nd…
tensorflower-gardener
committed

Commits on Nov 1, 2016

Move SparseTensor into its own framework.py, add get_shape().

ebrevdo
authored and
tensorflower-gardener
committed

Commits on Oct 29, 2016

Automated rollback of change 137564676
tensorflower-gardener
committed

Commits on Oct 28, 2016

Adding CPU kernels for tf.scatter_nd(), tf.scatter_nd_update(), tf.scatter_nd_add(), tf.scatter_nd_sub(), tf.scatter_nd_mul() and tf.scatter_nd_div() as well as gradient functions for tf.scatter_nd…
tensorflower-gardener
committed

Commits on Sep 29, 2016

Update trace, matrix_set_diag, matrix_diag_part and their gradients to work for rectangular matrices.
tensorflower-gardener
committed

Commits on Sep 21, 2016

Add SpaceToBatchND and BatchToSpaceND ops for > 2 spatial dimensions.
tensorflower-gardener
committed

Commits on Sep 13, 2016

1. Rename batch_matrix_* ops to matrix_*. Specifically, this applies to
tensorflower-gardener
committed

Commits on Sep 9, 2016

Rename NoGradient -> NotDifferentiable, to make it clear

Vijay Vasudevan
authored and
tensorflower-gardener
committed

Commits on Sep 2, 2016

Remove NoGradient for ops which could have gradients defined but don't.

Mrinal Kalakrishnan
authored and
tensorflower-gardener
committed

Commits on Aug 19, 2016

Fix gather gradient for empty slices

girving
authored and
tensorflower-gardener
committed

Commits on Aug 18, 2016

Updated the check_numerics function to also validate the gradient corresponding

benoitsteiner
authored and
tensorflower-gardener
committed
Automated rollback of change 130554783

girving
authored and
tensorflower-gardener
committed

Commits on Aug 17, 2016

Add a BUILD rule for extract_image_patches_grad_test, and make the code py3

gunan
authored and
tensorflower-gardener
committed
Fix gather for nonempty indices, empty slices

girving
authored and
tensorflower-gardener
committed

Commits on Aug 12, 2016

Merge changes from github.

benoitsteiner
authored and
tensorflower-gardener
committed