D4RL

Introduced by Justin Fu et al. in D4RL: Datasets for Deep Data-Driven Reinforcement Learning archive 2025-07-28

D4RL is a collection of environments for offline reinforcement learning. These environments include Maze2D, AntMaze, Adroit, Gym, Flow, FrankKitchen and CARLA.

Source: https://sites.google.com/view/d4rl/home Image Source: https://sites.google.com/view/d4rl/home

Homepage Benchmarks Papers JSON

Benchmarks archive 2025-07-28

All 2 leaderboards whose dataset resolves to this page shown (sort by any header). "First row" is the archive's own first row at snapshot, in the archive's row order; nothing here re-ranks and metric direction is not asserted.

		First row (archive order)	Paper	Code
D4RL	D4RL	PMDB Average Reward 88.2	Model-Based Offline Reinforcement Learning with...	huawei-noah/HEBO +2	9	Compare
Offline RL	D4RL	KFC Average Reward 81.8	Koopman Q-learning: Offline Reinforcement Learning via...	—	3	Compare

Papers archive 2025-07-28

10 shown of 10 papers with a leaderboard row on this dataset's benchmarks, newest first. The archive's own "papers using this dataset" list was never published, so this is the benchmark-backed subset; the archive's count for this dataset is 538. The Syntology column is from Syntology's graph (read 2026-09-24), stated per sample; it is not part of any archive number.

			Date	Samples run Syntology
Any-step Dynamics Model Improves Future Predictions for Online and Offline Reinforcement Learning	1	1	27 May 2024	ran 1 of 1 samples (0 unverified)
Primal-Attention: Self-attention through Asymmetric Kernel SVD in Primal Representation	1	1	31 May 2023	ran 1 of 1 samples (0 unverified; 1 pointer-only for licence)
Model-Based Offline Reinforcement Learning with Pessimism-Modulated Dynamics Belief	3	1	13 Oct 2022	not harvested
cosFormer: Rethinking Softmax in Attention	3	1	17 Feb 2022	not harvested
Flowformer: Linearizing Transformers with Conservation Flows	1	1	13 Feb 2022	ran 3 of 6 samples (3 unverified)
Koopman Q-learning: Offline Reinforcement Learning via Symmetries of Dynamics	0	2	2 Nov 2021	not harvested
Decision Transformer: Reinforcement Learning via Sequence Modeling	20	2	2 Jun 2021	ran 17 of 26 samples (9 unverified; 6 pointer-only for licence)
Rethinking Attention with Performers	7	1	30 Sep 2020	ran 9 of 16 samples (7 unverified; 6 pointer-only for licence)
Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention	8	1	29 Jun 2020	ran 3 of 8 samples (5 unverified; 3 pointer-only for licence)
Reformer: The Efficient Transformer	10	1	13 Jan 2020	ran 6 of 8 samples (2 unverified)

Dataset loaders archive 2025-07-28

www.tensorflow.org/datasets/catalog/d4rl_mujoco_walker2d tensorflow/datasetstfjax

www.tensorflow.org/datasets/catalog/d4rl_mujoco_hopper tensorflow/datasetstfjax

www.tensorflow.org/datasets/catalog/d4rl_mujoco_halfcheetah tensorflow/datasetstfjax

www.tensorflow.org/datasets/catalog/d4rl_mujoco_ant tensorflow/datasetstfjax

4 loaders as listed in the archive; links are outbound and not re-checked here.

Tasks archive 2025-07-28

Continuous Control Decision Making Offline RL D4RL Gym halfcheetah-random Gym halfcheetah-medium Gym halfcheetah-expert Gym halfcheetah-medium-expert Gym halfcheetah-medium-replay Gym halfcheetah-full-replay Adroid pen-human Adroid hammer-human Adroid door-human Adroid relocate-human Adroid pen-cloned Adroid hammer-cloned Adroid door-cloned Adroid relocate-cloned

License archive 2025-07-28

Apache-2 / CC-BY

Modalities archive 2025-07-28

Environment

Languages archive 2025-07-28

No language tagged.

Variants archive 2025-07-28

D4RL

1 variant name, as the archive lists them.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections