Dries Smit
Dries Smit
### Describe the benchmarking experiment/task The main goal of the Mava JAX release is to match the performance of [this](https://arxiv.org/abs/2103.01955) paper. Our notion page, documenting our progress, can be found...
### Please describe the purpose of the feature. Is it related to a problem? This feature would allow the critic network to utilise recurrent neural networks. Recurrent networks should improve...
### Feature Make the executor have an environment loop instead of the other way around. It makes more intuitive sense to run the executor instead of the environment loop (which...
### What do you want to investigate? Determine a better Mava standard that would make the executors and trainers more composable. We want to make it easy to add new...
### Describe the bug It seems that the record_every flag is broken. It waits for an initial number of runs as specified by record_every before displaying an episode. But after...
### Describe the bug When running in single process mode, Mava tries to run everything on the GPU. This causes problems when we run multiple tests on a machine with...
Hello there. My example code setup looks as follows: ``` import time import tensorflow as tf from absl.testing import absltest import launchpad as lp class learner(): def __init__(self) -> None:...
Only merge after #40
Only merge after #39
Only merge after #38