- 13 1月, 2017 25 次提交
-
-
由 David Majnemer 提交于
Change: 144371549
-
由 Jonathan Hseu 提交于
Reformatted core/platform/posix/port.cc as a test. Change: 144368934
-
由 A. Unique TensorFlower 提交于
Change: 144366312
-
由 Peter Hawkins 提交于
Change: 144362931
-
由 A. Unique TensorFlower 提交于
Change: 144360309
-
由 A. Unique TensorFlower 提交于
Change: 144356967
-
由 A. Unique TensorFlower 提交于
Change: 144355590
-
由 Gunhan Gulsoy 提交于
Change: 144354566
-
由 A. Unique TensorFlower 提交于
Change: 144354547
-
由 A. Unique TensorFlower 提交于
Change: 144354294
-
由 A. Unique TensorFlower 提交于
[XLA] Recognize any reduction where dimensions to keep are consecutive in memory as effective reduction to vector Change: 144353864
-
由 Mustafa Ispir 提交于
User code will look like as follows: opt = tf.SyncReplicasOptimizer(...) train_op = opt.minimize(total_loss, global_step=global_step) sync_rep_hook = opt.make_session_run_hook(is_chief) with training.MonitoredTrainingSession(master=master, is_chief=is_chief, hooks=[sync_rep_hook]) as mon_sess: while not mon_sess.should_stop(): mon_sess.run(training_op) Change: 144353039
-
由 A. Unique TensorFlower 提交于
Change: 144352405
-
由 David G. Andersen 提交于
Change: 144351555
-
由 Asim Shankar 提交于
Change the member variable to be of type Classifier interface instead of the implementation (TensorFlowImageClassifier) in the activity class. My intention is to create new Classifier implementations that use the TensorFlow Java API (org.tensorflow.Graph, Session etc.) instead of the Android contrib API (org.tensorflow.contrib.android.TensorFlowInferenceInterface). This re-organization will make the switch between Classifier implementations easier during testing. While at it, some minor cleanups: - Get rid of unnecessary "throws IOException" - Use a factory function instead of an initializer function. Change: 144348772
-
由 David G. Andersen 提交于
Change: 144344623
-
由 A. Unique TensorFlower 提交于
Fix bug in shape inference for set operations. It errantly had been setting first dim of output indices and values to first dim of inputs. Add logic to infer the output rank of DenseToSparse ops from the 2nd (sparse) input arg, even if the 1st (dense) arg's rank is unknown. Replace explicit rank checks with InferenceContext.WithRank* and InferenceContext.WithValue. Change: 144344168
-
由 Patrick Nguyen 提交于
This also exposes meta_Graph_pb2.TensorInfo as tf.TensorInfo. Change: 144344131
-
由 Sergio Guadarrama 提交于
Change: 144342278
-
由 A. Unique TensorFlower 提交于
Change: 144341695
-
由 Shanqing Cai 提交于
The wrapper generates debug-dump directories inside a session_root directory, with each directory corresponding to a Session.run() call. In addition to the usual GraphDef file and dumped Tensor Event files, each run directory contains special files: "_tfdbg_run_fetches_info" and "_tfdbg_run_feed_keys_info", containing a str representation of the fetches and feed_dict.keys() used during the Session.run() call, respectively. The debug-dump directories have the naming pattern of "run_<epoch_timestamp>_<uuid>". Change: 144340734
-
由 Shanqing Cai 提交于
Change: 144340064
-
由 Joshua V. Dillon 提交于
transformation of iid Student's t-distributions and should not be confused with the Multivate Student's T-distribution (which is an example of an https://en.wikipedia.org/wiki/Elliptical_distribution). Change: 144338393
-
由 A. Unique TensorFlower 提交于
Change: 144331488
-
由 A. Unique TensorFlower 提交于
Only enables for gradient descent so far. Change: 144331045
-
- 12 1月, 2017 15 次提交
-
-
由 Asim Shankar 提交于
Make the C library and JNI shared library targets Android friendly by linking with the smaller android runtime when building with --config=android Relatedly, strip all but the JNI symbols from libtensorflow_jni.so (regardless of build configuration) to trim its size down (by roughly 50%). Link in the Java libraries in the Android examples. The longer term intention is to encourage use of the TensorFlow Java API in Android and do away with the TensorFlowInferenceInterface class (and related JNI code) currently used in Android. This will provide a single, more thoroughly tested API for use in all Java settings - Android or not. An update to the Android example to switch to this will come in a follow up change. NOTES: - For expediency of this change, the C API call: TF_LoadSessionFromSavedModel is not available when building for Android. I will look into fixing that separately. - Linking in the JNI library required by the TensorFlow Java API results in a small increase (0.7%) in binary size of libtensorflow_demo.so An unrelatedly, rename libtensorflow-jni.so to libtensorflow_jni.so to be consistent with other shared libraries created in tensorflow. Change: 144320074
-
由 A. Unique TensorFlower 提交于
Change: 144289849
-
由 A. Unique TensorFlower 提交于
Change: 144289263
-
由 Asim Shankar 提交于
Fixes #6799 Change: 144288761
-
由 A. Unique TensorFlower 提交于
Change: 144284257
-
由 A. Unique TensorFlower 提交于
Change: 144283085
-
由 A. Unique TensorFlower 提交于
Change: 144282810
-
由 A. Unique TensorFlower 提交于
Change: 144282152
-
由 A. Unique TensorFlower 提交于
to the NodeDef representation. Change: 144281952
-
由 Andrew Selle 提交于
Change: 144279869
-
由 Vijay Vasudevan 提交于
Change: 144278086
-
由 Neal Wu 提交于
Change: 144275556
-
由 A. Unique TensorFlower 提交于
Change: 144273322
-
由 A. Unique TensorFlower 提交于
Change: 144273220
-
由 A. Unique TensorFlower 提交于
Change: 144271336
-