- 11 3月, 2017 8 次提交
-
-
由 Stephan Hoyer 提交于
This is faster, conceptually simpler and avoids finicky issues with dtype precision. Change: 149782530
-
由 A. Unique TensorFlower 提交于
V2 Checkpoints are common, now the default from all TFX pipelines. Users of this tool, when filling the 'file_name' flag are likely to submit an actual filename, rather than the required prefix. This error message should catch those cases where the user is very likely to have meant the prefix. Change: 149781964
-
由 Shanqing Cai 提交于
CL/149672538 has made TensorHandles directly feedable for Session.run() calls. This eliminates the need to call TensorHandles.eval() to get the value of a Tensor and feed it back in during tfdbg NodeStepper's cont() (i.e., continue-to) calls. This change improves the memory efficiency and performance of NodeStepper.cont() by eliminating unnecessary Tensor-numpy and numpy-Tensor copying. Change: 149769853
-
由 Kiril Gorovoy 提交于
Test using this classification signature using the Predict API. Change: 149769547
-
由 A. Unique TensorFlower 提交于
Change: 149769532
-
由 A. Unique TensorFlower 提交于
initializes global variables. Change: 149769121
-
由 Jonathan Hseu 提交于
Change: 149768595
-
由 A. Unique TensorFlower 提交于
Add some additional values args to name_scope calls. Change: 149765769
-
- 10 3月, 2017 22 次提交
-
-
由 Christopher Olston 提交于
Rename the new TryConcat()/TrySplit() to Concat()/Split() and mark the existing Concat()/Split() as deprecated. This is the first change in a sequence that will remove the crashy Concat()/Split(). Change: 149755537
-
由 Frank Perbet 提交于
graph_editor: move match.py into the the test folder as it is not meant to be part of the graph_editor public API. Change: 149741516
-
由 Blake Hechtman 提交于
Change: 149732502
-
由 Sukriti Ramesh 提交于
Change: 149720034
-
由 A. Unique TensorFlower 提交于
Change: 149714094
-
由 Mustafa Ispir 提交于
Change: 149706759
-
由 A. Unique TensorFlower 提交于
Add a sample program to evaluate clock cycle profiler and support clock cycle profiling on android 64bit Change: 149704577
-
由 Brennan Saeta 提交于
In order to support end-to-end debugging and performance profiling tooling for the TensorFlow::XLA toolchain, this change adds a DebugMetadata proto to the HloInstruction class, and pipes it through the tf2xla stack. Change: 149703349
-
由 A. Unique TensorFlower 提交于
Refactored the debugger plugin test to use a real server. Modified the plugin code itself appropriately. This makes the test more real and more streamlined since we do not need to fake various objects any more. Furthermore, this refactoring sets the stage for testing endpoints that read directly from disk and bypass the events multiplexer (which we want to do in order to read health pills at individual steps). Change: 149702774
-
由 A. Unique TensorFlower 提交于
Change: 149699250
-
由 A. Unique TensorFlower 提交于
do a deep copy of a BucketizedColumn as we use tuple(sanitized_boundaries). Change: 149698536
-
由 A. Unique TensorFlower 提交于
Change: 149695936
-
由 Zakaria Haque 提交于
Makes DNNLinearCombinedEstimator public. Leaves alias for the private name which will be removed once all call sites are updated. Change: 149683187
-
由 A. Unique TensorFlower 提交于
Change: 149681364
-
由 Justine Tunney 提交于
This CL works around what appears to be a bug in how Bazel sandbox works. If an sh_binary() is listed under tools in genrule(), then the contents of the data attribute of that sh_binary() do not seem to be a part of the sandbox file system. Change: 149678142
-
由 Justin Lebar 提交于
Change: 149674904
-
由 A. Unique TensorFlower 提交于
Change: 149673098
-
由 Shanqing Cai 提交于
This is motivated by, among other goals, the need to enhance memory efficiency during TFDBG's stepper operations. The stepper caches TensorHandles to already-continued-to tensors and use them as feeds if later continue-to actions depend on the tensors as transitive inputs. However, previously the TensorHandles had to be converted to Numpy arrays by calling eval() and the Numpy arrays were then fed back to next Session.run() calls. This mode of operation involved at least two unnecessary tensor-numpy and numpy-tensor copying. This CL makes it possible to use the ResourceHandle representations TensorHandles directly as feed values, eliminating the need for the aforementioned copying. To this end, the following changes are made 1) the underlying representations of TensorHandles are changed from string to ResourceHandle. A custom numpy struct type is created to allow ResourceHandle of the TensorHandle subtype to be fed during Session.run() calls. 2) added GetSessionHandleOpV2, which deprecates GetSessionHandleOp. The V2 op outputs a DT_RESOURCE Tensor, instead of a string Tensor in the deprecated version. Change: 149672538
-
由 A. Unique TensorFlower 提交于
consistency. Change: 149672185
-
由 Derek Murray 提交于
Change: 149671298
-
由 Jianwei Xie 提交于
Change: 149658231
-
由 A. Unique TensorFlower 提交于
Delete AUC metrics from MultiClassHead. This removed AUC metrics from DNNClassifier, LinearClassifier and DNNLinearCombinedClassifier when n_classes > 2. Change: 149656850
-
- 09 3月, 2017 10 次提交
-
-
由 James Qin 提交于
Also fixed typo: "obsfucate" -> "obfuscate" Change: 149627297
-
由 A. Unique TensorFlower 提交于
Change: 149617825
-
由 A. Unique TensorFlower 提交于
Change: 149615375
-
由 Sukriti Ramesh 提交于
Change: 149615173
-
由 Zakaria Haque 提交于
Makes the head/multi-head API public and updates selected users while leaving other users using aliases. This is CL#1 of a series of CLs to make head/multi-head API public and migrate all users. This CL, + Makes the Head interface and factory functions public. + Updates all tf-learn internal and SIR usage. + Leaves aliases for the legacy private names which will be removed with all existing usages in next CLs. + Also, updates documentation. Change: 149613397
-
由 Mark McDonald 提交于
This should be enough to get the Android examples working directly through Android Studio, that is: check out the GitHub repo, set up bazel location, load build.gradle in AS and run. Change: 149609160
-
由 A. Unique TensorFlower 提交于
Change: 149606199
-
由 A. Unique TensorFlower 提交于
Change: 149590447
-
由 Dandelion Mané 提交于
The EventMultiplexer and EventAccumulator classes are fairly tightly tied to TensorBoard, so we should migrate them to the tensorboard/ folder in preparation for extracting TensorBoard from the TensorFlow repo. Change: 149587771
-
由 Dandelion Mané 提交于
Change: 149580266
-