Status of the project?

[le-dawg]

Hello,

how would you describe the current status? I am failing at getting anywhere with the tf.adanet due to a shitty documentation.

[ffs97]

Hi @le-dawg,

Thanks for your interest. We haven't been working on this for some time as this was a course project that we collaborated on. @baldvaritesh and I have been thinking of trying to build this into a production level api with proper integration with pytorch. We will have to check the licenses and stuff for that but should not be an issue.

For now, you could clone the repo and play around with it. The code is in running status and the adanet algorithm works smoothly. However, I wouldn't put it in any production level stuff since there might be a couple of differences from the original paper as they decided to skip out some lower level details in the paper and we didn't look at the tf code too well.

We will also work on documenting the code and the usage but given it's pytorch, the code is super readable and surprisingly not too complex. I will update you as and when we pan out the project. Leaving this issue open for now.

[le-dawg]

OK, sounds interesting. I have been stalled for over a day to understand how to get going with time series and the required data transofrmation for tf.estimators. A real pity that no adanet.keras api exists that would allom me to skip building input_fn.

[ffs97]

It would be interesting to work with time series models using Adanet. Hopefully we can bring the project to production stage soon enough. Will keep you posted.