Metaworld

Metaworld copied to clipboard

Really High successful rate for RL2-PPO in ML45

1

Thanks for offering us this wonderful environments and benchmarks. When I checked the new benchmark results of ML10/ML45 on the updated paper https://arxiv.org/pdf/1910.10897.pdf, I found that the new RL2-PPO performance...

waterhorse1

Why it does not success even when the score is high?

2

I tested some environment with different task settings. For example on reach-v2, I think it is strange that it does not success even when the score is 4700+. I tried...

alexxchen

In the V2 push-wall, pick-place-wall, push-back, and shelf-place environments, setting `rand_init` to False and then resetting the environment would lead to the following error when calling `step()`: ``` Traceback (most...

zchuning

The lower bound for the random reset space of the object in disassemble is higher than the upper bound for the first index. This appears to be an issue for...

ottofabian

Currently, the observation space does not specify a dtype, which means it defaults to float32, however the returned numpy array is by default in float64. This will lead to exceptions...

ottofabian

can not render under Tesla-V100 GPU

2

Hi, it seems that the same bug always exists under V100-32G GPU environment: ``` Found 0 GPUs for rendering. Using device 0. File "mjsim.pyx", line 156, in mujoco_py.cymj.MjSim.render File "mjsim.pyx",...

zhengsipeng

Replacing mujoco-py with pybind11-based mujoco bindings

3

Recently, other big benchmark suites, such as [OpenAI gym](https://github.com/openai/gym/blob/024b0f5160716b01effc27de45615bac7620e8b6/README.md?plain=1#L51-L52) and [DMC](https://github.com/deepmind/dm_control/blob/41d0c7383153f9ca6c12f8e865ef5e73a98759bd/README.md?plain=1#L61-L68), switched to the new [pybind11-based mujoco bindings](https://pypi.org/project/mujoco/). Since these bindings are made available in DeepMind's official mujoco repository, I...

ottofabian

Seeding not working for v2 environments

2

I am trying to get deterministic behavior for the same seed> However, running the following piece of code prints about different result every time. ``` import metaworld ml10 = metaworld.ML10(seed=0)...

buoyancy99

State random vector uses global RNG instead of per-environment RNG

3

I was trying to use your environments as standalone gym tasks without the metalearning aspect. While I do not think this issue conflicts with the design choices made for the...

ottofabian

mujoco_py.cymj.GlfwError: Failed to initialize GLFW

1

When I use this repo on the server without interface, I Get an error : mujoco_py.cymj.GlfwError: Failed to initialize GLFW, I know it is because the MJViewer, but how can...

RayYoh