HugeCTR

HugeCTR copied to clipboard

[Question] Low throughtput with our testing data.

3

Hi HugeCTR team, I'm testing the HugeCTR performance with our data of real scenario, but the throughtput is unecpected low. I double checked our data and scrpits, but still no...

AtekiRyu

The current [Movie-Lens example](https://github.com/NVIDIA-Merlin/HugeCTR/blob/master/notebooks/movie-lens-example.ipynb) uses KNN to make a inference demo but produces suspicious results which have a lot of queries returning the same recommendation. According to @vinhngx, there is...

jershi425

[Question] why can not use multi gpu of a example notebook

8

when i try Embedding Training Cache Example [https://nvidia-merlin.github.io/HugeCTR/master/notebooks/embedding_training_cache_example.html](url) if i use multi gpu by set vvgpu to [[1,2]], i wil get error at second pass : HugeCTR Version:3.6 =====================================================Model Fit=====================================================...

LarryZhangy

[Question] Failed to preprocess criteo data

2

when i try to run a sample of wdl https://github.com/NVIDIA-Merlin/HugeCTR/tree/master/samples/wdl i got a error at step :Preprocess the Dataset Through NVTabular bash preprocess.sh 0 criteo_data nvt 1 0 1 #...

LarryZhangy

[BUG] notebooks of hugectr_wdl_prediction can not run

2

**Describe the bug** when i try notebook https://nvidia-merlin.github.io/HugeCTR/master/notebooks/hugectr_wdl_prediction.html i got error when run preprocess.py TypeError: Unsupported type: Cannot convert object of type to Node. **To Reproduce** Steps to reproduce the...

LarryZhangy

[BUG] Installing of sparse_operation_kit from pip failed

5

**Describe the bug** A clear and concise description of what the bug is. **To Reproduce** Steps to reproduce the behavior: 1. Use AMI https://aws.amazon.com/releasenotes/deep-learning-ami-gpu-tensorflow-2-9-ubuntu-20-04/ to spin a ec2 cluster instance...

silpara

[BUG] hps_demo.ipynb crashes with core dump

3

**Describe the bug** When running the hps_demo.ipynb notebook, with the latest docker container 22.05, the third code cell where we train the model will crash with a core dump. I...

PeterDykas

A process in the process pool was terminated abruptly while the future was running or pending

3

As the title, I run the script: python ../tools/embedding_plugin/performance_profile/preprocess.py --src_csv_path=train.txt --dst_csv_path=train.out.txt --normalize_dense=0 --feature_cross=0 in the official docker，but it core dumped with A process in the process pool was terminated abruptly...

lizhen2017

[Question] Complete guide to train wdl model using all 1TB criteo data in only six minutes

1

I know the performance of HugeCTR + NVTabular is awesome, the Criteo 1TB Click Logs Dataset can be processed and a full model can be trained in only six minutes....

LarryZhangy

v3.7 preview document update

1

zehuanw