dyth

Here's some `gdb` debugging information, courtesy of some suggestions from @obilaniu. The [backtrace](https://github.com/deepmind/dm_control/files/8401040/backtrace.txt) and ``` (gdb) x/i $pc => 0x7fffe827dea5 : mov (%r12),%esi (gdb) p/x $r12 $1 = 0x0 ```...

Thanks, @saran-t, for your fast reply. Do you think it's possible that `dm_control.suite` is somehow overwriting the `mujoco-py` environment registrations during its import and wiping the pointer?

I also think the issue might also be with `mujoco-py`. FWIW, there is [a link to the OpenAI slack in this codebase](https://github.com/robintyh1/onpolicybaselines/blob/58d401622b38a9127437cf48b1b62b27028d694a/onpolicyalgos/discrete_actions_space/trpo_ordinal/run_trpo_ordinal.py#L3) that says import mujoco-py first. To that end,...

I was hoping to do RL from scratch of a 56D humanoid from states. I saw that [the 2019 parameters weren't intended for training from scratch](https://github.com/deepmind/dm_control/issues/155#issuecomment-694558468), so I thought I'd...

Thanks for the information -- is this action preprocessor different from action normalization (ie just linearly rescaling the action space to [-1, 1])?

@tomhennigan sorry for tagging, but might there be an update?

I followed the discussion in https://github.com/google/jax/issues/7430 to come up with the following, but it seems quite hacky. ``` import haiku as hk import jax import jax.numpy as jnp def f(x):...

Hi there, the version of gym minigrid I used was `af0e00fe906fccfc70ef116a5ee89751e1441942` and the BabyAI version was `863f3529371ba45ef0148a48b48f5ae6e61e06cc`

Summarised transcript of an internal discussion between @dyth and @sunchipsster1: * the gifs in the first post are transposed and the the x and y axes should be swapped *...

Hello! Apologies for not getting back to you sooner. I'm not too familiar with the BabyAI codebase at this point but AFAIK attempts to make a bot for MiniGrid envs...