gntoni

Make sure to change: ` model.eval()` for: ``` model.train() optimizer.zero_grad() ``` in: https://github.com/lijiaman/CASENet/blob/0aa475d856ebaa5fa7bbb1a144da606c9ad4198c/train_val/model_play.py#L28

I am also very interested in this. [Here](https://docs.wandb.ai/guides/runs/grouping) are some examples in WandB. I would be awesome to have something like that.

@avaziri > Some users report that installing in this way prevents interoperability with gym, or causes segfaults. Those interoperability issues were before the last version of gym was released, now...

I agree with Julien, it is confusing to keep seeing warnings after setting the log severity level for a session.

Any advances on this? I'm having the same error on ubuntu 14.04 using a conda environment. I tried to run setup() build --verbose, but the output is the same: running...