aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
* Test for distributed (grpc) runtime in OSS TensorFlowGravatar A. Unique TensorFlower2016-03-18
| | | | | | | | | | | | | | | | | | | | See README.md for detailed descriptions of the usage of the tools and tests in this changeset. Three modes of testing are supported: 1) Launch a local Kubernetes (k8s) cluster and run the test suites on it (See local_test.sh) 2) Launch a remote k8s cluster on Google Container Engine (GKE) and run the test suite on it (See remote_test.sh) 3) Run the test suite on an existing k8s TensorFlow cluster (Also see remote_test.sh) Take the remote test for example, the following steps are performed: 1) Builds a Docker image with gcloud and Kubernetes tools, and the latest TensorFlow pip installed (see Dockerfile) 2) Launches a Docker container based on the said image (see test_distributed.sh) 3) From within the image, authenticate the gcloud user (with credentials files mapped from outside the container), configer the k8s cluster and launch a new k8s container cluster for TensorFlow workers 4) Generate a k8s (yaml) config file and user this yaml file to create a TensorFlow worker cluster consisting of a certian number of parameter servers (ps) and workers. The workers are exposed as external services with public IPs (see dist_test.sh) 5) Run a simple softmax MNIST model on multiple workers, with the model weights and biases located on the ps nodes. Train the models in parallel and observe the final validation cross entropy (see dist_mnist_test.sh) Change: 117543657
* Creating a build target for kernels in contrib.Gravatar A. Unique TensorFlower2016-03-18
| | | | Change: 117523151
* Fix a static_cast overflow in WorkSharder Shard impl.Gravatar A. Unique TensorFlower2016-03-18
| | | | Change: 117520810
* Rollback of "TensorFlow: move eigen some NN code from our third_party/eigen3 ↵Gravatar Vijay Vasudevan2016-03-18
| | | | | | | copy to being part of TF, add tests." Change: 117519243
* Fix two potential asynchrony bounds-check bugs in transpose op.Gravatar David G. Andersen2016-03-18
| | | | Change: 117518926
* Refactoring common checking and size computation code into aGravatar David G. Andersen2016-03-18
| | | | | | | | | | | separate struct that is shared by all of the image resizers. Normalizes the error checking across all of the resizers. Also added a max size check to nearest_neighbor - because of the floats, it starts to produce bad results after 2^24px in either direction. Not that anyone does that, but it's good to be precise about it. Change: 117516271
* Add optional functionality to save StepStats proto to the filesystem in the ↵Gravatar Andrew Harp2016-03-18
| | | | | | | Android demo, for performance analysis. Enable by hardcoding kSaveStepStats to true or passing "--copt -DSAVE_STEP_STATS" to bazel build. Change: 117512949
* TensorFlow: move eigen some NN code from our third_party/eigen3 copyGravatar Vijay Vasudevan2016-03-18
| | | | | to being part of TF, add tests. Change: 117509710
* TensorFlow: update eigen to latest release that has a fix to too large frame.Gravatar Vijay Vasudevan2016-03-18
| | | | Change: 117506296
* Add fast path for identity transpose.Gravatar A. Unique TensorFlower2016-03-18
| | | | Change: 117505457
* Removed height:0px on tf-multi-checkbox because it was behaving poorly for ↵Gravatar A. Unique TensorFlower2016-03-18
| | | | | | embedded components. Change: 117504934
* Update losses in contrib.layers.Gravatar A. Unique TensorFlower2016-03-18
| | | | Change: 117504830
* Patching placer to print out a list of devices when a placement cannot be ↵Gravatar A. Unique TensorFlower2016-03-18
| | | | | | made to aid debugging. This instantly answers the question: did I specify it wrong or is the device not found? Change: 117493711
* Added basic support for float16 on CPUs and older GPUs.Gravatar Benoit Steiner2016-03-18
| | | | | Also fixed compilation issues with cuda devices that support the compute model 5.3 Change: 117493644
* GPU implementations of space_to_depth and depth_to_space.Gravatar A. Unique TensorFlower2016-03-18
| | | | Change: 117493386
* C++ gradients: Changes API to require specification of node outputs (also ↵Gravatar A. Unique TensorFlower2016-03-18
| | | | | | added some code cleanup). Change: 117488572
* Make EmbeddingWrapper accept an explicit embedding_size argument so it does ↵Gravatar Lukasz Kaiser2016-03-18
| | | | | | not rely on cell.input_size. Change: 117484994
* Clarifying comment and adding check in should_stop after hitting a confusing ↵Gravatar A. Unique TensorFlower2016-03-18
| | | | | | snag. Change: 117484454
* Adding a library for creating a stable categorical color scale for ↵Gravatar A. Unique TensorFlower2016-03-18
| | | | | | real-time, changing data. Change: 117483893
* Fix qualified name of SummaryWriter in htmlGravatar Geoffrey Irving2016-03-18
| | | | Change: 117483092
* Added check for 0 length input before it can get to Eigen.Gravatar Martin Wicke2016-03-18
| | | | Change: 117482953
* fix implicit int64 to int32 conversionGravatar Jianmin Chen2016-03-18
| | | | Change: 117475266
* fix doc of enqueue_manyGravatar A. Unique TensorFlower2016-03-18
| | | | Change: 117471008
* fix indices in image describing dynamic_stitchGravatar A. Unique TensorFlower2016-03-18
| | | | Change: 117456435
* Move Relu and SoftMax Op declarations into header files so they can be ↵Gravatar A. Unique TensorFlower2016-03-18
| | | | | | | | | registered by experimental devices. Right now tensorflow/core/kernels explicitly depends on all Eigen devices that might want to implement any of the templated Eigen Ops. This is because the template classes that need to be specialized are defined in .cc files, so the specializations themselves have to appear there too. Moving the classes to .h files allows us to use arbitrary Eigen devices defined outside of tensorflow/core, which fits better with the intent behind core/kernels. Over time more kernels may need to be refactored this way for the same reason. Change: 117452814
* Add padding between main, in-extract and out-extract graphs.Gravatar Dan Smilkov2016-03-18
| | | | Change: 117444098
* Typo controling -> controlling.Gravatar A. Unique TensorFlower2016-03-18
| | | | Change: 117420208
* Allowing users to optionally provide a global step tensor and name to SDCA's ↵Gravatar A. Unique TensorFlower2016-03-18
| | | | | | minimize method. Change: 117401811
* Fixing a test for posix environment for memmapped regionsGravatar A. Unique TensorFlower2016-03-16
| | | | Change: 117394750
* Move Relu Op declarations into relu_op.h.Gravatar A. Unique TensorFlower2016-03-16
| | | | Change: 117392494
* Move Relu Op declarations into relu_op.h.Gravatar A. Unique TensorFlower2016-03-16
| | | | Change: 117387358
* Forcing copy of input tensor, switching to use FastBoundsCheckGravatar David G. Andersen2016-03-16
| | | | Change: 117384840
* Simplify pylint disables.Gravatar Josh Levenberg2016-03-16
| | | | Change: 117384554
* Because TensorFlow knows the entire graph of your computations, itGravatar A. Unique TensorFlower2016-03-16
| | | | | | | | can automatically use the [backpropagation algorithm](http://colah.github.io/posts/2015-08-Backprop/) to efficiently determine how your variables affect the cost you ask it to minimize. Change: 117382393
* Rollforward of "Merge changes from github."Gravatar Geoffrey Irving2016-03-16
| | | | Change: 117375570
* Another bugfix in run_and_gather_logs.Gravatar Eugene Brevdo2016-03-16
| | | | Change: 117375431
* Allows for importing of graphs with no variable nodes.Gravatar A. Unique TensorFlower2016-03-16
| | | | Change: 117365836
* Fixing an int64->int32 conversion that may have made concatoffset notGravatar David G. Andersen2016-03-16
| | | | | | | work with (ridiculously) large tensors. Also switching an existing test to use FastBoundsCheck to increase consistency with where we're taking the other ops. Change: 117364143
* Change dep list of tf_gpu_kernel_library rule.Gravatar A. Unique TensorFlower2016-03-16
| | | | Change: 117361209
* Extending tensorflow::Env with support for memmapped filesGravatar A. Unique TensorFlower2016-03-16
| | | | Change: 117356696
* Eliminating the DoValidation/DoCompute split for training kernels andGravatar David G. Andersen2016-03-16
| | | | | | | | | eliminating several other asynchrony-related validation bugs. (This turns out to also shrink the code a bit by eliminating a fair bit of redundant input accessor code between the validation and compute phases). Haven't benchmarked, but if anything, this change should be performance-positive. Change: 117353014
* Fix another unused code warning.Gravatar Josh Levenberg2016-03-16
| | | | Change: 117347879
* C++ gradients: split functions gradient code out into its own file.Gravatar A. Unique TensorFlower2016-03-16
| | | | Change: 117346884
* Adds scalar_logistic_loss() and apply_regularization() to contribGravatar Wei Ho2016-03-16
| | | | Change: 117305077
* Rollback of: "Merge changes from github."Gravatar Vijay Vasudevan2016-03-16
| | | | Change: 117304114
* Add tf.contrib.lookup.string_to_index and tf.contrib.lookup.index_to_string ↵Gravatar Yutaka Leon2016-03-16
| | | | | | to map strings to IDs and viceversa. Change: 117303981
* - Add a generic immutable hash table op that is initialized once and used to ↵Gravatar Yutaka Leon2016-03-16
| | | | | | | | | | | | | | | | | | | map key tensors to values. - Add a table initializer given the keys and values as tensors. Example use case: keys = tf.constant([0, 1], tf.int64) values = tf.constant(["hello", "world"]) default_value = "UNK" table = tf.contrib.lookup.HashTable(tf.contrib.lookup.KeyValueTensorInitializer(keys, values), default_value) input_tensor = tf.constant([0, 2], tf.int64) out = table.lookup(input_tensor). tf.initialze_all_tables().run() # or table.init.run() print out.eval() # Returns ["hello", "UNK"] Change: 117301920
* Merge changes from github.Gravatar Martin Wicke2016-03-16
| | | | Change: 117301677
* Disable //third_party/tensorflow/core/distributed_runtime:master_test.Gravatar Derek Murray2016-03-16
| | | | | | | | | This test is flaky. We need a more robust solution for starting multiple processes and connecting them; `PickUnusedPortOrDie()` is not robust to multiple concurrent tests on the same host. A possible approach would be to add a portserver, or add a handshake between the parent test process and the server processes. Change: 117293607
* fix error message formatting in session_manager.pyGravatar A. Unique TensorFlower2016-03-16
| | | | Change: 117285646
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-03 14:01:03 +0000