Test/validation data for Semantic Policy

[yinoue0426]

I've downloaded the data for training the semantic policy (here), but it seems that it mostly only contains the training data. It does contain the "tests_unseen" directory, but only 25 samples are available.

Since I want to evaluate the semantic policy model, would it be possible to share the full evaluation data, i.e. train_unseen, train_seen, val_unseen, and val_seen? If that is difficult, would it be possible to share the code used to generate the data?

[soyeonm]

Hi, I can't guarantee but I will work on it and get back to you in a few days. Thanks!

[yinoue0426]

Sounds great, thanks so much!