xtra-computing / g3 Goto Github PK

Summary:
I am getting this error when running gcn demo sample with the dataset pubmed:
"terminate called after throwing an instance of 'std::invalid_argument'"

Reproducibility:
./gcn --feature_file=../../dataset/gcn/pubmed.svmlight --graph_file=../../dataset/gcn/pubmed.graph --split_file =../../dataset/gcn/pubmed.split

To Reproduce:
1 gpu with 1GB RAM

For example:

1. Clone gunrock
2. Switch to the dev-refactor branch

Make sure to:
$ git submodule init
$ git submodule update
etc.
etc.

Expected Behavior:
A clear and concise description of what you expected to happen.

Actual Results:
A clear and concise description of what you actually experienced.

System Configuration:

GPU: GTX 970
CPU: Intel(R) Xeon(R) CPU E3-1280 v5 @ 3.70GHz
OS: Ubuntu 18.04.1 LTS
Cuda compilation tools, release 10.0, V10.0.130
NVIDIA-SMI 415.13
Driver Version: 415.13
CUDA Version: 10.0
gcc (Ubuntu 6.4.0-17ubuntu1) 6.4.0 20180424

Feel free to create pull-requests if you get a chance, and I can review and merge them. And then you can add gunrock/gunrock as an external git submodule and it will be maintained for you (obviously, as we do updates to it).

Hi,

I opened issue #6 and asked about the Running "Case-study-COVID-19-in-Singapore" demo using G3. You stated that the code base of the Covid-19 demo on G3 is hard-coded, and the LSTM part is not 100% functional. Therefore it is not pushed to this repo yet.

However, I have another question related to this. I have a dataset containing some time-series data on a GNN. I want to train a GNN using the G3 for the prediction over this time-series.

For example, you can consider a dataset of Covid-19 cases. So, based on this, I want to train 10 sequential days of confirmed cases in 5 cities to predict the 11th day of confirmed cases in these cities. So my question is that, is it possible to use time-series data to train a GNN in current the current G3?

I reviewed the sample data you put in this repo for the G3 (G3/dataset/gcn/). None of them are time-series data and time-series prediction. Regarding the current API of G3 and the lack of the LSTM module in it, I want to know whether we can train a model for time series prediction using G3 or not.

Thanks for your time. :)

Hi, I reviewed the "Case-study-COVID-19-in-Singapore" demo and executed it successfully on my system. However, I did not understand the relation between this demo and the G3 program. Here are my questions:

1- The "Case-study-COVID-19-in-Singapore" demo uses PyTorch LSTM and Linear modules to train the described 3-layered GNN module. So, to use G3 for the training, how do we change this defined PyThorh module to a G3 one?

2- How do we change the input data of this demo to one suitable for G3? As far as I realized, the datasets in the web-app demo are somehow different from the datasets in this demo.

3- In sum, implementing this demo with G3 is unclear. Therefore, I would appreciate that if you clarify this.

Thanks.