- 15 8月, 2017 2 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 165195082
-
由 A. Unique TensorFlower 提交于
1. There is a bug in in deriving the formulas for batch norm gradient calculation. This commit corrects the formula and updates the implementation. 2. Learning from mistake, this commit uses gradient_checker to test batch_norm_grad in a more generic way -- this method calculates the "real" gradients by evaluating the graph twice, and then compare it with the gradient. RELNOTES: Update formula for fused batch norm in xla PiperOrigin-RevId: 165190486
-
- 14 8月, 2017 1 次提交
-
-
由 A. Unique TensorFlower 提交于
to reduce memory consumption. On the wire, this should be fully compatible unless values > uint32max were actually ever used, which is unlikely. PiperOrigin-RevId: 165171764
-
- 13 8月, 2017 2 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 165078268
-
由 A. Unique TensorFlower 提交于
wrapped in the generated code. PiperOrigin-RevId: 165078237
-
- 12 8月, 2017 26 次提交
-
-
由 Shanqing Cai 提交于
PiperOrigin-RevId: 165053647
-
由 Benoit Steiner 提交于
PiperOrigin-RevId: 165049643
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 165049580
-
由 Ali Yahya 提交于
PiperOrigin-RevId: 165049398
-
由 Adam Roberts 提交于
Allow the next input in ScheduledOutputTrainingHelper to be computed by any callable by removing requirement that it be a Layer. PiperOrigin-RevId: 165042108
-
由 Francois Chollet 提交于
PiperOrigin-RevId: 165039827
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 165035012
-
由 Ali Yahya 提交于
PiperOrigin-RevId: 165034457
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 165034131
-
由 A. Unique TensorFlower 提交于
used with the channels_first data_format. PiperOrigin-RevId: 165028212
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 165026447
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 165026308
-
由 Pete Warden 提交于
Adds a basic training script for a simple audio model to our examples. See third_party/docs_src/tutorials/audio_recognition.md for full documentation PiperOrigin-RevId: 165025732
-
由 Adam Roberts 提交于
Allow specification of sample_id_shape and sample_id_dtype in seq2seq.BasicDecoder and add a new InferenceHelper. PiperOrigin-RevId: 165019969
-
由 Alexandre Passos 提交于
PiperOrigin-RevId: 165019327
-
由 Alexandre Passos 提交于
New bindings show the actual default values for attrs on the generated code. This will lead to nicer documentation. PiperOrigin-RevId: 165018499
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 165015675
-
由 Eugene Brevdo 提交于
Add unit tests for stability of the sorts. PiperOrigin-RevId: 165015637
-
由 Andrew Harp 提交于
PiperOrigin-RevId: 165013681
-
由 A. Unique TensorFlower 提交于
Permit multiple REGISTER_XLA_OP calls for the same Op, as long as they register the Op on disjoint sets of devices. PiperOrigin-RevId: 165010633
-
由 Shanqing Cai 提交于
PiperOrigin-RevId: 165007814
-
由 Vijay Vasudevan 提交于
PiperOrigin-RevId: 165003359
-
由 Adam Roberts 提交于
PiperOrigin-RevId: 165001965
-
由 Andrew Harp 提交于
benchmark_model: add optional inter-benchmark delay and max benchmark length params to help reduce effects of thermal throttling and standardize run-length between graphs. PiperOrigin-RevId: 164996768
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 164989700
-
由 Shanqing Cai 提交于
Previously, a grpc-gated debug op has two modes: DISABLED and ENABLED. This CL splits the ENABLED state into two states: READ_ONLY and READ_WRITE. * READ_ONLY is equivalent to the previous ENABLED state, wherein a debug op publishes debug tensor to the grpc debug server and proceeds. It can be regarded as a "watchpoint" that doesn't block execution. * READ_WRITE is a "breakpoint". In addition to publishing the debug tensor, it blocks and awaits a EventReply proto response from the grpc debug server before proceeding. PiperOrigin-RevId: 164987725
-
- 11 8月, 2017 9 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 164943597
-
由 Kay Zhu 提交于
compute constant backend. PiperOrigin-RevId: 164940970
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 164939527
-
由 A. Unique TensorFlower 提交于
Merge code from PR #11940 with internal changes from cl/164796436, and update Python tests to also run on GPU. PiperOrigin-RevId: 164929133
-
由 Anna R 提交于
description for Benchmarks page to Community index page. PiperOrigin-RevId: 164924906
-
由 Mark Heffernan 提交于
As part of this change make HloAliasAnalysis a thinner layer which basically only holds a map from HloValue to HloBuffer and vice versa. PiperOrigin-RevId: 164923041
-
由 A. Unique TensorFlower 提交于
164923041 by meheff: Make HloAliasAnalysis updatable after changes to the HLO graph. As part of this change make HloAliasAnalysis a thinner layer which basically only holds a map from HloValue to HloBuffer and vice versa. -- PiperOrigin-RevId: 164923041
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 164920220
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 164917891
-