- 13 12月, 2018 20 次提交
-
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 225253270
-
由 Jian Li 提交于
PiperOrigin-RevId: 225249344
-
由 Allen Lavoie 提交于
Copies and pastes the existing Optimizer checkpointing code, and stops adding unconditional dependencies on slot variables (which were based on ops.uid() and so not reproducible across program runs). PiperOrigin-RevId: 225248820
-
由 Katherine Wu 提交于
PiperOrigin-RevId: 225245412
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 225237733
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 225236744
-
由 Francois Chollet 提交于
PiperOrigin-RevId: 225231668
-
由 A. Unique TensorFlower 提交于
custom op, it is up to the tf-lite user to provide the implementation. Best to assume it exists so the user can implement. PiperOrigin-RevId: 225228337
-
由 Dan Moldovan 提交于
Reduce the cost of serializing ConversionOptions to code, by using a more efficient inspect.util.getqualifiedname, reducing its max_depth and falling back to caching the value in the namespace. The latter step makes it more difficult to run the generated code afterwards, but it should in turn speed up the conversion process. This also adds an extra check to tf_decorator to improve robustness. PiperOrigin-RevId: 225226256
-
由 A. Unique TensorFlower 提交于
These tests share the same assertion: that weighting a particular class's loss over other classes (by passing in `sample_weight` into `model.fit`) leads to a lower evaluation loss when evaluating test data limited to that class compared to evaluating all test data. My theory is that the models in these tests are not trained enough for that assumption to always hold true, which is why they are flaky. Increased the weight from 2 to 10 and the training epochs from 5 to 10. PiperOrigin-RevId: 225218063
-
由 Zhenyu Tan 提交于
PiperOrigin-RevId: 225217785
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 225212001
-
由 Gaurav Jain 提交于
PiperOrigin-RevId: 225210711
-
由 Artem Belevich 提交于
PiperOrigin-RevId: 225208397
-
由 Peter Hawkins 提交于
PiperOrigin-RevId: 225205868
-
由 A. Unique TensorFlower 提交于
1. Only MIN_COMBINED mode is supported; 2. Reshape the output to [d0,..., dn * unpack_size] if input shape is [d0, ..., dn]. 3. Only uint32 is supported for the input; 4. Output data type is bfloat16; 5. Only uint8 or uint16 is supported for the original unpacked input. PiperOrigin-RevId: 225203930
-
由 Skye Wanderman-Milne 提交于
PiperOrigin-RevId: 225202451
-
由 Scott Zhu 提交于
This contrain was originally added due to the different weights format issue between canonical and cudnn (extra input bias). Now since the input bias is feeded as zeros in cudnn mode, and weights are unified into one format. Having bias regularizer should not be a issue. PiperOrigin-RevId: 225193782
-
由 Mihai Maruseac 提交于
PiperOrigin-RevId: 225189182
-
由 Skye Wanderman-Milne 提交于
Removing the LoopCond of a while_loop can cause the partitioner to fail with: A cross-device loop must have a pivot predicate For some reason this only triggers with while_v2 (the lowered while loop is slightly different than what would be produced by the original while_loop). PiperOrigin-RevId: 225188075
-
- 12 12月, 2018 20 次提交
-
-
由 Gaurav Jain 提交于
In addition, fix a few eval() calls as well as remove some @test_util.run_v1_only annotations. PiperOrigin-RevId: 225180248
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 225178266
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 225140840
-
由 A. Unique TensorFlower 提交于
Format in the previous state didn't give the timezone. PiperOrigin-RevId: 225138116
-
由 A. Unique TensorFlower 提交于
Remove :android_tensorflow_lib_selective_registration* aliases, targets using selective registration can now use the :android_tensorflow_lib_lite* targets. PiperOrigin-RevId: 225134497
-
由 Skye Wanderman-Milne 提交于
PiperOrigin-RevId: 225131361
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 225131218
-
由 Li, Guizi 提交于
Conflicts: tensorflow/core/graph/mkl_layout_pass.cc
-
由 Sanjoy Das 提交于
PiperOrigin-RevId: 225127595
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 225125955
-
由 Mark Heffernan 提交于
No functional change. Rename the proto message Layout to LayoutProto, and Tile to TileProto. Define in-place replacement C++ classes named Layout and Tile with an interface which mirrors the protobuf generated code interface. Having these data structures as C++ classes enables greater flexibility in the interface, enables enforcement of invariants, and potential performance improvements. PiperOrigin-RevId: 225121052
-
由 A. Unique TensorFlower 提交于
RELNOTES: Add an ignore_unknown argument to parse_values which suppresses ValueError for unknown hyperparameter types. Such hyperparameter are ignored. parse_values('a=1,b=foo', {a: int}) Raises a ValueError parse_values('a=1,b=foo', {a: int}, ignore_unknown=True) does not raise a ValueError, and returns {'a': 1} PiperOrigin-RevId: 225117666
-
由 A. Unique TensorFlower 提交于
(1) Able to trace tensors when the model is executed on the CPU. (previously, it could only trace when the model is executed on TPU) (2) Allow the user to specify the op-names and op-types to be excluded or included for tracing via regular expressions. (3) Two new trace modes: (1) tracing the vector norm of the tensor and (2) tracing the maximum of the absolute values of all elements in the tensor. (4) Attach the replica-ID to a traced tensor value so that the post-processing tool (Tensor-Inspector) can reconstruct the whole tensor from all replicas. (5) An API to trace tensors programmatically. (6) Allow writing the trace to stdout (previously, it must be written to a file). PiperOrigin-RevId: 225112219
-
由 Smit Hinsu 提交于
Co-Authored-By: Ntrevor-m <tmorris@nvidia.com>
-
由 Yifei Feng 提交于
PiperOrigin-RevId: 225110993
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 225110815
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 225110544
-
由 Andy Ly 提交于
PiperOrigin-RevId: 225109110
-
由 Dan Moldovan 提交于
PiperOrigin-RevId: 225107801
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 225107483
-