FrozenLake
A sparse-reward grid navigation reinforcement learning benchmark environment in which an agent must navigate from a start point to a goal on a frozen lake grid. Its sparse reward signal makes it a standard testbed for evaluating credit assignment in reinforcement learning algorithms.
Timeline (last 90 days)
ViSkill achieves an overall success rate of 0.89 on the three environments Sokoban, FrozenLake, and PrimitiveSkill, rising to 0.91 when cold start is enabled.
Sokoban tests long-horizon planning and spatial deduction, FrozenLake focuses on decision-making to avoid traps in stochastic environments, and PrimitiveSkill focuses on the reuse of basic skills.
The authors evaluated RLDS on four agent benchmarks: FrozenLake, HotpotQA, ScienceWorld, and DeepResearch.
RLDS achieved an improvement of +11.5 points on ScienceWorld (95% confidence interval [+9.8, +13.3]) and +9.8 points on FrozenLake ([+7.0, +12.8]).
Applying VQC to classic RL benchmark environments such as CartPole and FrozenLake has been covered in multiple papers available for reproduction and comparison.
All Facts (5)
ViSkill achieves an overall success rate of 0.89 on the three environments Sokoban, FrozenLake, and PrimitiveSkill, rising to 0.91 when cold start is enabled.
50%UnverifiedSokoban tests long-horizon planning and spatial deduction, FrozenLake focuses on decision-making to avoid traps in stochastic environments, and PrimitiveSkill focuses on the reuse of basic skills.
50%UnverifiedThe authors evaluated RLDS on four agent benchmarks: FrozenLake, HotpotQA, ScienceWorld, and DeepResearch.
50%UnverifiedRLDS achieved an improvement of +11.5 points on ScienceWorld (95% confidence interval [+9.8, +13.3]) and +9.8 points on FrozenLake ([+7.0, +12.8]).
50%UnverifiedApplying VQC to classic RL benchmark environments such as CartPole and FrozenLake has been covered in multiple papers available for reproduction and comparison.
50%