Jonathan Shen
Jonathan Shen
I kept both the explicit initialization in the constructor and the lazy initialization in createIODescriptors. The explicit initialization is needed so that the threads share the same tensor - if...
Removed lazy initialization of iSize based on discussion above and added some hacky code to convert. It seems to still work fine based on my tests on AlexNet. Please let...
R: @charlesccychen
Sorry - yes that is a good rephrasing.
Thanks, it was the only choice from https://github.com/apache/beam/blob/master/sdks/python/apache_beam/typehints/OWNERS maybe that should be updated.
I think that's likely some kind of OOM, but I have no idea why it would just quit without printing any kind of error. If it runs fine with --run_locally=cpu...
This has to do with our custom implementation of https://github.com/tensorflow/lingvo/blob/master/lingvo/core/recurrent.py which processes the tensors one timestep at a time. Conversion is very straightforward ```python # Assuming padding is [max_sequence_length, batch_size]...
The env looks fine. Is this in sync mode or async mode? I checked our run it was in async mode with 8 machines x 2 gpus and got 0.9...
When you update lingvo version you almost always will need to update tf-nightly version as well.