SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2010.14484 · NeurIPS · 2020

One Solution is Not All You Need: Few-Shot Extrapolation via Structured MaxEnt RL

Saurabh Kumar, Sergey Levine, Chelsea Finn, Aviral Kumar

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 7 functions out of this paper's own repositories and ran 5 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
egiob/diversityisallyouneed-sb3 pwc_unofficial 5 of 7
FunctionStatusWhere it lives
crawler_disc_on_y_legs Ran egiob/diversityisallyouneed-sb3/stable_baselines3/common/behaviour_functions.py
code served (permissive licence) · get_code("9f397819ac46d7f7")
crawler_is_flying Ran egiob/diversityisallyouneed-sb3/stable_baselines3/common/behaviour_functions.py
code served (permissive licence) · get_code("d832171813a56c1f")
crawler_sensor_outputs Ran egiob/diversityisallyouneed-sb3/stable_baselines3/common/behaviour_functions.py
code served (permissive licence) · get_code("82e6460bcd964905")
create_mlp Ran egiob/diversityisallyouneed-sb3/stable_baselines3/common/torch_layers.py
code served (permissive licence) · get_code("ef103f787c48beca")
sum_independent_dims Ran egiob/diversityisallyouneed-sb3/stable_baselines3/common/distributions.py
code served (permissive licence) · get_code("ee85c29e20e06d17")
get_actor_critic_arch Not yet run egiob/diversityisallyouneed-sb3/stable_baselines3/common/torch_layers.py
code served (permissive licence) · get_code("f2ded9c22d608f44")
kl_divergence Not yet run egiob/diversityisallyouneed-sb3/stable_baselines3/common/distributions.py
code served (permissive licence) · get_code("df19469790ae30d9")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

While reinforcement learning algorithms can learn effective policies for complex tasks, these policies are often brittle to even minor task variations, especially when variations are not explicitly provided during training. One natural approach to this problem is to train agents with manually specified variation in the training task or environment. However, this may be infeasible in practical situations, either because making perturbations is not possible, or because it is unclear how to choose suitable perturbation strategies without sacrificing performance. The key insight of this work is that learning diverse behaviors for accomplishing a task can directly lead to behavior that generalizes to varying environments, without needing to perform explicit perturbations during training. By identifying multiple solutions for the task in a single environment during training, our approach can generalize to new situations by abandoning solutions that are no longer effective and adopting those that are. We theoretically characterize a robustness set of environments that arises from our algorithm and empirically find that our diversity-driven approach can extrapolate to various changes in the environment and task.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2010.14484")
get_code_for_paper("2010.14484")
have("2010.14484")

Connect an agent — have() is free.