Dong Meng

[BUG]when exporting hugeCTR model, NVT model is hardcoded to run on CPU

2

**Describe the bug** see https://github.com/NVIDIA-Merlin/NVTabular/blob/72f2fe4e882e126856dd96054807acbc929f64e3/nvtabular/inference/triton/__init__.py#L427-L428 **Steps/Code to reproduce bug** NA **Expected behavior** NVT backend should be allowed to run on GPU **Environment details (please complete the following information):** NA

[BUG] cuDF failure processing full Criteo dataset, when the parquet files are exported from big query

2

**Describe the bug** cuDF failure processing full Criteo dataset, when the parquet files were exported from big query. CSV and parquet from GCS works. **Steps/Code to reproduce bug** https://gist.github.com/mengdong/d6a24fc266d9806ccd74cd9890b67c6a replace...

OpRegularizerManager could not handle ops

7

Hello, I have tried a few examples from tensorflow/model with morphnet (lenet and resnet), a simple mnist model (https://github.com/mengdong/morph-net/blob/master/morph_net/examples/mnist/mnist-tutorial.py) works. However, I ran into problems in some other more complex...

add automatic mixed precision

1

Automatic Mixed Precision for Tensorflow has been recently introduced: https://medium.com/tensorflow/automatic-mixed-precision-in-tensorflow-for-faster-ai-training-on-nvidia-gpus-6033234b2540 This PR adds automatic mixed precision to self attention GAN training. We've tested speed/convergence impact on V100/T4. We have seen...