Name: GeForce GTX 780 Ti major: 3 minor: 5 memor圜lockRate(GHz): 1.15 07:50:30.090327: I tensorflow/core/common_runtime/gpu/gpu_:1618] Found device 0 with properties: 07:50:29.987980: I tensorflow/stream_executor/platform/default/dso_:44] Successfully opened dynamic library nvcuda.dll 07:50:25.464075: I tensorflow/stream_executor/platform/default/dso_:44] Successfully opened dynamic library cudart64_100.dll Try to provide a reproducible test case that is the bare minimum necessary to generate the problem.

If including tracebacks, please include the full traceback. Include any logs or source code that would be helpful to diagnose the problem.

Print('Number of devices: '.format(time.time() - start_time)) Source code / logs Strategy = tf.distribute.MirroredStrategy() Model_pile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=) Model_normal.add(tf.(units=10, activation='softmax')) I've uninstalled and re-installed 'pip install tensorflow-gpu' twice. The train on the CPU works well, but then when moving to do it with the distributed strategy shows the error below. Be sure to convey here why it's a bug in TensorFlow or a feature request.

Exact command to reproduce: model_distributed.fit(X_train, y_train, epochs=4, batch_size=25)ĭescribe the problem clearly here.

GPU model and memory: 4x GTX780ti 3GB each.

GCC/Compiler version (if compiling from source): N/A.

Bazel version (if compiling from source): N/A.

TensorFlow version (use command below): 2.0.0.

TensorFlow installed from (source or binary): binary.

iPhone 8, Pixel 2, Samsung Galaxy) if the issue happens on mobile device:

OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Win10 1903.

Have I written custom code (as opposed to using a stock example script provided in TensorFlow): Custom code/ but copy from the stock example (nothing out of the ordinary).

Vspd locomotive