mbrl-lib
mbrl-lib copied to clipboard
Library for Model Based RL
# 🚀 Feature Request Current implementation only works with pixel observations. Adding support for low-dimensional observations (i.e., the standard gym obs) should be relatively easy, as it shouldn't require more...
# 🚀 Feature Request The [planet visualizer](https://github.com/facebookresearch/mbrl-lib/blob/main/mbrl/diagnostics/planet_visualizer.py) is an ad-hoc version used during the debugging stages. Now that the PlaNet implementation is stable, we should update it to be given...
# 🚀 Feature Request Add a batch dimension to all operations in CEM so that it can optimize several problems at once. ## Motivation The current implementation of our CEM...
Hi, currently reward_fn is independent from environment class (mbrl.models.ModelEnv) and accepts as input actions and next observation. In practice more general, dependent on environment parameters reward functions are needed. For...
Hi, Usually in practice we have some dynamic model which at least approximate the true environment x_{t+1} = f(x_t, u_t) (e.g from Newton laws). My question is how to include...
# 🚀 Feature Request PILCO: A Model-Based and Data-Efficient Approach to Policy Search. - http://mlg.eng.cam.ac.uk/pub/pdf/DeiRas11.pdf - PILCO is SOTA for small d systems and would be great to be able...
# 🚀 Feature Request A configuration option to save all data for every trial. ## Motivation When debugging it can be very useful to see the intricate details of how...
# 🚀 Feature Request Create Replay Buffer Iterators that can select training and validation data in various "interesting" ways, similar to `TransitionIterator` and `BootStrapIterator` in [https://github.com/facebookresearch/mbrl-lib/blob/b0aabd79941efe8b56bcabbd1b43bf497b9b1746/mbrl/replay_buffer.py](https://github.com/facebookresearch/mbrl-lib/blob/b0aabd79941efe8b56bcabbd1b43bf497b9b1746/mbrl/replay_buffer.py) Examples: 1) Select transitions...
Heya, I think this not a great idea, happy to discuss why: https://github.com/facebookresearch/mbrl-lib/blob/c7a7425d6ac74d781cececc5c440eab87f4b7c67/mbrl/algorithms/pets.py#L24
# 🚀 Feature Request Add functionality to upload dynamics models /policies to the HF hug at end of training or during training for sharing / fine-tuning. This would like like...