Upstream changes from internal #2028
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Also update typings in package.json and add typing dependency on sinon (its provided by wct) Change: 119988943
…te the inverse of the adjoint of a matrix. This is convenient in various gradient computations where it also saves an explicit transpose op. Implements gradients for batch_matrix_determinant and optimizes implementation of gradient for matrix_determinant. Change: 119990836
try_lock is used by eigen3/unsupported/Eigen/CXX11/src/Tensor/TensorRunQueue.h which is disabled by default, so it left unnoticed. TESTED: - passed opensource_build Change: 119991000
The autoreload logic is factored out into a behavior, and is tested. Change: 119991467
Change: 119992758
Change: 119992878
Auto-reloading defaults on and is controlled by a setting. To have a place to put that setting, also add a settings panel. The settings panel uses paper-dialog and is opened by clicking a settings icon button in the global top right. Minor changes to how the toolbar is laid out by CSS to ensure that the settings all show up in a line and don't bunch up together. Also, the "TensorBoard" message on the left no longer takes a fixed pixel width (everything is flex), so the toolbar will use space more appropriately when the window size is small. Change: 119997005
…elected indices had a 0. Making it emulate the old behavior because gather stopped working for length 0 Tensors. e.g.: Gathering row 1 from a Tensor of shape [2, 0] should result in a [1, 0] Tensor, but instead resulted in an error like: AttributeError: index 1 is not in [0, 2) Change: 119998790
…locator that has zero memory in its heap. Change: 120002729
Change: 120004793
Change: 120008243
…ession in the supervisor. Make it innocuous to start a queue runner multiple times. Change: 120010279
Change: 120010519
Also clarify that google.any expansion is not supported. Change: 120013039
…scan. Change: 120017883
Usage: `tf.sparse_add(sp_tensor, tensor)`; CPU only for now.
The newly introduced ScatterNdFunctor can later be extended to handle
ScatterNd{Update,Add,Sub} ops, similar to the existing GatherNd op.
Change: 120018023
Change: 120018234
Change: 120018303
This change takes advantage of the fact that a dense gradient tensor must have the same shape as the corresponding input to the forward-pass op. It may enable additional optimizations (such as the ability to use the accumulate_n aggregator) that rely on static shape information. This change revealed some broken code, for which fixes are included: * The shape function for `tf.diag_part()` incorrectly used an equality test on shapes, rather than a compatibility test. * The gradient function for `tf.nn.top_k()` incorrectly returned a length-1 vector for the gradient w.r.t. `k`, which it should have been a scalar. * A unit test in `control_flow_ops_py_test` inconsistently used vectors and scalars. Change: 120038251
…encies, probably done in pure python TF, rather than wrapped C++. Starting by moving lbeta from math_ops to special_math_ops, since a bugfix requires it to depend on control_flow_ops. The bug was: lbeta([]) returned -inf, whereas it should have returned 0.0. Change: 120069192
Change: 120069280
Change: 120076873
// OLD Benchmark Time(ns) CPU(ns) Iterations ------------------------------------------------------------------------ BM_ConvFloatDepthwiseBkFilterCPU1_conv0 281152179 280588497 100 588.2M items/s 32_112_112_3_8_24_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv1 760242956 758694909 100 580.1M items/s 32_112_112_64_1_64_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv2 383554418 382741182 100 574.9M items/s 32_56_56_128_1_128_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv3 98924384 98665676 100 557.2M items/s 32_56_56_128_1_128_3_3_2_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv4 94237506 94005920 100 585.0M items/s 32_28_28_128_1_128_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv5 106895864 106648144 100 515.7M items/s 32_14_14_512_1_512_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv6 69247718 69078442 100 398.0M items/s 32_7_7_1024_1_1024_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv7 70304661 70126053 100 588.1M items/s 32_112_112_3_8_24_3_3_2_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv8 67619710 67447142 100 611.4M items/s 32_112_112_3_8_24_3_3_2_1_cpu1 // NEW 1-thread Benchmark Time(ns) CPU(ns) Iterations ------------------------------------------------------------------------ BM_ConvFloatDepthwiseBkFilterCPU1_conv0 59981294 59569328 100 2.7G items/s 32_112_112_3_8_24_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv1 165631344 165250674 100 2.6G items/s 32_112_112_64_1_64_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv2 76910026 76705735 100 2.8G items/s 32_56_56_128_1_128_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv3 21491439 21375872 100 2.5G items/s 32_56_56_128_1_128_3_3_2_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv4 18677714 18587209 100 2.9G items/s 32_28_28_128_1_128_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv5 2347423 23377934 100 2.3G items/s 32_14_14_512_1_512_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv6 17066829 16982791 100 1.6G items/s 32_7_7_1024_1_1024_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv7 14822571 14744419 100 2.7G items/s 32_112_112_3_8_24_3_3_2_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv8 14325480 14254559 100 2.8G items/s 32_112_112_3_8_24_3_3_2_1_cpu1 // NEW 4-threads Benchmark Time(ns) CPU(ns) Iterations ------------------------------------------------------------------------ BM_ConvFloatDepthwiseBkFilterCPU4_conv0 21809044 69141049 100 7.4G items/s 32_112_112_3_8_24_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv1 57704422 192333505 100 7.5G items/s 32_112_112_64_1_64_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv2 29761264 91848609 100 7.2G items/s 32_56_56_128_1_128_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv3 9075773 26429821 100 5.9G items/s 32_56_56_128_1_128_3_3_2_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv4 7276754 22100190 100 7.4G items/s 32_28_28_128_1_128_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv5 6756189 24510067 100 8.0G items/s 32_14_14_512_1_512_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv6 4837993 17723279 142 5.6G items/s 32_7_7_1024_1_1024_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv7 6676347 19935585 100 6.0G items/s 32_112_112_3_8_24_3_3_2_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv8 5951583 17181079 100 6.8G items/s 32_112_112_3_8_24_3_3_2_1_cpu4 TESTED: - passed opensource_build - passed unit tests Change: 120125325
Benchmark Base (ns) New (ns) Improvement ------------------------------------------------------------------ BM_MaxPool_32_112_112_64_3_3_2_VALID_1 28173747 28956041 -2.8% BM_MaxPool_32_56_56_192_3_3_2_VALID_1 14467716 14581478 -0.8% BM_MaxPool_32_28_28_352_3_3_2_VALID_1 5318842 5367336 -0.9% BM_MaxPool_32_14_14_576_3_3_2_VALID_1 1331917 1351642 -1.5% BM_MaxPool_32_112_112_64_3_3_2_SAME_1 28757024 29005280 -0.9% BM_MaxPool_32_56_56_192_3_3_2_SAME_1 15119295 15478783 -2.4% BM_MaxPool_32_28_28_352_3_3_2_SAME_1 5802450 5871220 -1.2% BM_MaxPool_32_14_14_576_3_3_2_SAME_1 1632582 1662128 -1.8% BM_MaxPool_32_112_112_64_3_3_2_VALID_4 28579650 8240771 +71.2% BM_MaxPool_32_56_56_192_3_3_2_VALID_4 14621344 4373595 +70.1% BM_MaxPool_32_28_28_352_3_3_2_VALID_4 5404303 1571711 +70.9% BM_MaxPool_32_14_14_576_3_3_2_VALID_4 1343607 427873 +68.2% BM_MaxPool_32_112_112_64_3_3_2_SAME_4 29195151 8204002 +71.9% BM_MaxPool_32_56_56_192_3_3_2_SAME_4 15314088 4642979 +69.7% BM_MaxPool_32_28_28_352_3_3_2_SAME_4 6094918 1777112 +70.8% BM_MaxPool_32_14_14_576_3_3_2_SAME_4 1643584 544554 +66.9% TESTED: - passed opensource_build - passed unit tests Change: 120128184
…d an utility to convert a freezed graph into this format. Change: 120128412
its inputs. Change: 120128592
- remove flags (pushed to internal client) - thread-parallel execution for CountExtremeleyRandomStats op. - critical time-seeding fix. Change: 120129936
…s, softmax_cross_entropy_loss, and sum_of_pairwise_squares_loss. Refactoring old losses for consistency. Change: 120130871
Change: 120136109
Class represents multi-indexed batches of Dirichlet Multinomial distributions. Initialized with parameters alpha, which broadcast to arbitrary shapes to match arguments in e.g. dist.pdf(x). Change: 120138028
Change: 120139351
Change: 120140181
…and expanding documentation of get_collection. Change: 120141411
Change: 120142925
TESTED: - passed opensource_build (only known failing test loss_ops test fails) - passed unit tests Change: 120156989
Change: 120161004
Change: 120164016
Fixes a bug in gather that would segfault when gathering from very large (>2^31 entry) parameter tensors. Gather can now handle index vectors with more than 2^31 entries, (if you have enough memory). Change: 120171737
Change: 120172505
Combines the 6 entities (sparse, dense) x (features, weights, delta-weights)
into a single class (FeaturesAndWeights) which tracks all features and
weights, hiding their underlying representation. This class is built by
composing other classes:
FeaturesAndWeights
SparseFeaturesAndWeights
(examples_by_group_)
WeightsAndDeltas
WeightsByGroup
(delta_weights_by_group_)
DenseFeaturesAndWeights
(features_by_group_)
WeightsAndDeltas (same as above)
Also adds a microbenchmark.
Change: 120173207
Change: 120183287
Change: 120184745
Change: 120185825
Change: 120186451
Change: 120186712
Change: 120192253
Change: 120192816
- Oneof values should be printed even when equal to the default. - Fields should be printed in tag number order, not declaration order. Change: 120223509
Change: 120231603
|
We found a Contributor License Agreement for you (the sender of this pull request), but were unable to find agreements for the commit author(s). If you authored these, maybe you used a different email address in the git commits than was used to sign the CLA (login here to double check)? If these were authored by someone else, then they will need to sign a CLA as well, and confirm that they're okay with these being contributed to Google. |
fsx950223
pushed a commit
to fsx950223/tensorflow
that referenced
this pull request
Nov 28, 2023
…ime-docker Add new runtime image for TF
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.