-
Hi.
After implementing pre-recorded expert data with the size of 1e6. I have realized that `np.random.choice` with `replace=Fasle` is extremely slow to the point of unusable. (batch size 100)
I a…
-
Hi,
I have read your excellent work and try to reproduce it. But,I was little confused about the results of cql baseline.
In **hopper-medium-v0** and **hopper-medium-replay-v0**(mixed), I got the…
-
Implement [TD3+BC](https://arxiv.org/pdf/2106.06860.pdf) with the [unplugged-rl](https://github.com/deepmind/deepmind-research/tree/master/rl_unplugged) dataset. For this get a MuJoCo student license,…
-
### What is the problem?
I get the following error when reproducing cql.
```
gym.error.Error: Attempted to look up malformed environment ID: b'7/site-packages/ray/rllib/d4rl'. (Currently all …
-
Although @Mobius1D has done some work on it recently in https://github.com/JuliaReinforcementLearning/ReinforcementLearning.jl/pull/384 , it seems to deviate from what I thought. So I'd write down som…
-
**Describe the bug**
Running both `./scripts/test` and `./scripts/lint` on master give errors.
**To Reproduce**
Following the instructions in `CONTRIBUTING.md`:
```
./scripts/test
+ FLAG…
-
Hi,
Thanks for sharing this interesting work. However, I have few questions in the paper:
1) Could I know how to get the score of an expert policy that you used to normalize the score in Tables?…
-
Python version - 3.7.5
Libs versions -
tensorboard==2.1.1
tensorboard-plugin-wit==1.6.0.post2
tensorflow-addons==0.7.1
tensorflow-datasets==4.3.0
tensorflow-estimator==2.1.0
tensorflow-gpu==2…
-
Making publicly available datasets usable directly for offline reinforcement learning is crucial for RLDatasets.jl. Recently we have added support for d4rl datasets and d4rl-pybullet datasets.
Now …
-
I run some experiment in D4RL while I cannot reproduce results in several tasks.
For example,
in Halfcheetah-medium-expert-v0, I got
```
Epochs,Eval_Reward,Eval_Cvar,Eval_Std,d4rl scores
0,652…