adaptive-transformers-in-rl

adaptive-transformers-in-rl copied to clipboard

There seems to be a lot of bugs in the code especially the padding and end of sequence indexing. Can you please update the repo with bug free code used...

doltonfernandes

Regarding logic for first done indexes

1

Hi, Thanks for the code and the paper on using adaptive attention span in RL. In `train.py`, I haven't understood the logic for calculating `ind_first_done` in following line: https://github.com/jerrodparker20/adaptive-transformers-in-rl/blob/6f75366b78998fb1d8755acd2d851c461c82ee75/train.py#L1240 ....

victor-psiori

Stable Transformer on Pong

2

Hello, I am currently unable to recreate the results of the stable transformer on the Pong environment. I believe from the paper the last 100 episode returns should be ~17.62...

furmans

Hey, in the paper StablizingTransformer..., there is a gate unit in the moudle of the GTrxl, but in your default params, the --use_gate is False. Why?

weihongwei0586

Is this algorithm suitable for off-policy policy?

1

I just finished reading your paper, and I notice that it is an on policy method. And I wondering if anyone has tested it with an rl method that has...

dbsxdbsx

Then we can do some test on it.

alimai