| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
| |
The extra spaces were confusing bash's string-line-continuation from
the backslash `\` on the previous line.
PiperOrigin-RevId: 215964853
|
|
|
|
|
|
| |
`MapAndBatchDataset` whose user-provided functions have the property that each output argument take its value directly from an input argument (e.g. `lambda x, y: y, x`). This specialization can produce the result without having to schedule the function using the executor.
PiperOrigin-RevId: 215957592
|
|
|
|
| |
PiperOrigin-RevId: 215957535
|
|
|
|
| |
PiperOrigin-RevId: 215957327
|
|
|
|
|
|
| |
Enable GPU tests for cond_v2.
PiperOrigin-RevId: 215956220
|
|
|
|
| |
PiperOrigin-RevId: 215951354
|
|
|
|
| |
PiperOrigin-RevId: 215950376
|
|
|
|
| |
PiperOrigin-RevId: 215950207
|
|
|
|
| |
PiperOrigin-RevId: 215948571
|
|\
| |
| |
| | |
PiperOrigin-RevId: 215947463
|
| |
| |
| |
| |
| |
| | |
the model.
PiperOrigin-RevId: 215947354
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215946205
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215944829
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
preserve operand layouts.
Add an std::function member to the HloVerifier for a backend to specify the
function object used to determine whether an instruction can change layouts.
Use the function object to find out the non-layout-changing instructions and
check that such instructions should produce results with the same layouts as
its operands.
Add test cases.
PiperOrigin-RevId: 215941282
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215939542
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215937618
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215935319
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215930596
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215928419
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
attr values that are not overridden e.g. transpose_a in the matmul op).
This is required for backward compatibility (a binary built via an older version
of TF should still run on a newer version of TF, where some ops may have added
attrs).
For non-eager graph building, the default attr values of graph ops are added by
tensorflow::AddDefaultsToNodeDef().
We ran into this issue when running the same S4TF test cases via eager APIs --
some tests failed due to "missing attrs", but are fixed by this patch.
PiperOrigin-RevId: 215927271
|
| |
| |
| |
| |
| |
| |
| |
| | |
the optimizer isn't set, e.g. loading weights and then predict.
- Add load_weights for `KerasTpuModel`.
PiperOrigin-RevId: 215920993
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215917470
|
| |
| |
| |
| |
| |
| |
| |
| | |
Automated rollback of PR #21945
END_PUBLIC
Automated rollback of commit 863f61412fcc654840c6b67473b742ea4e5e964e. Revert #21945.
PiperOrigin-RevId: 215913175
|
| |
| |
| |
| |
| |
| |
| | |
Note to users manually patching ptxas from a later toolkit version:
Building NCCL requires the same version of ptxas and nvlink.
PiperOrigin-RevId: 215911973
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215910400
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
library.
Change XlaBinaryOp::Broadcast to use the BroadcastTo lowering, since it produces fewer extraneous reshapes and transposes. Even if the reshapes and transposes would later optimize away, this yields more readable output and makes life easier for HLO rewrites that run early.
Change in preparation for removing reshapes from SoftmaxCrossEntropyWithLogits.
PiperOrigin-RevId: 215906847
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215905026
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215874612
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215870962
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215857772
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215853105
|
| |
| |
| |
| | |
PiperOrigin-RevId: 215844450
|
| |
| |
| |
| |
| |
| | |
tf.train.init_from_checkpoint can be supported.
PiperOrigin-RevId: 215843249
|
| |
| |
| |
| |
| |
| |
| |
| | |
This was completely broken for CUDA versions > 9 and resulted in spurious warnings.
Reported in #22706#issuecomment-426861394 -- thank you!
PiperOrigin-RevId: 215841354
|
| |
| |
| |
| |
| |
| | |
The implementation is based on: https://openreview.net/pdf?id=S1dIzvclg.
PiperOrigin-RevId: 215824867
|
|\ \
| | |
| | |
| | | |
PiperOrigin-RevId: 215824410
|
| | |
| | |
| | |
| | |
| | |
| | | |
An environment variable (TF_EAGER_ENABLE_SMALL_TENSOR_CPU_PINNING) is provided to turn this off if necessary (its on by default).
PiperOrigin-RevId: 215821915
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215819072
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215814732
|
| | |
| | |
| | |
| | |
| | |
| | | |
to execute a different-sized matrix multiply than the eager tests.
PiperOrigin-RevId: 215814346
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215813259
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
was flawed. Added better test coverage.
Also added a extra test for a related symbolic shape inference operation that I first suspected to be broken.
PiperOrigin-RevId: 215812753
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215811680
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
* Add kernels for TensorListReserve. EmptyTensorList, TensorListElementShape, TensorListPushBack, TensorlistPopBack;
* Treat list type pretty much identical to Stack in the bridge for now;
* Support variant output by treating variant like a uint8 and leaving the interpretation up to the XlaExpression (variant type does not support tensor_data());
PiperOrigin-RevId: 215809335
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215808649
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215806953
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215802845
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215801897
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215801773
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 215797256
|