We lifted 9 functions out of this paper's own repositories and ran 7 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.
| Repository | Role | Ran |
|---|---|---|
| brucegeli/tce_rl | canonical | 7 of 9 |
| Function | Status | Where it lives |
|---|---|---|
| dir_go_up | Ran | brucegeli/tce_rl/mprl/util/util_file.py pointer only (licence: GPL-3.0) · get_code("92e5211355eecee6") |
| from_string_to_array | Ran | brucegeli/tce_rl/mprl/util/util_data_structure.py pointer only (licence: GPL-3.0) · get_code("2716f92c20b5f303") |
| get_dataset_dir | Ran | brucegeli/tce_rl/mprl/util/util_file.py pointer only (licence: GPL-3.0) · get_code("791d15c2f8feff67") |
| git_repos_old_vs_new | Ran | brucegeli/tce_rl/mprl/util/util_git.py pointer only (licence: GPL-3.0) · get_code("20f320988be5d87f") |
| make_iterable | Ran | brucegeli/tce_rl/mprl/util/util_data_structure.py pointer only (licence: GPL-3.0) · get_code("cdb63a3d0d6500be") |
| parse_dtype_device | Ran | brucegeli/tce_rl/mprl/util/util_data_structure.py pointer only (licence: GPL-3.0) · get_code("d628a55b9fcdd727") |
| remove_file_dir | Ran | brucegeli/tce_rl/mprl/util/util_file.py pointer only (licence: GPL-3.0) · get_code("2989d2eb82d281fe") |
| how_fast | Not yet run | brucegeli/tce_rl/mprl/util/util_debug.py pointer only (licence: GPL-3.0) · get_code("9c8d0df3d467a118") |
| run_time_test | Not yet run | brucegeli/tce_rl/mprl/util/util_debug.py pointer only (licence: GPL-3.0) · get_code("bac2eb5a8cec567f") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Current advancements in reinforcement learning (RL) have predominantly focused on learning step-based policies that generate actions for each perceived state. While these methods efficiently leverage step information from environmental interaction, they often ignore the temporal correlation between actions, resulting in inefficient exploration and unsmooth trajectories that are challenging to implement on real hardware. Episodic RL (ERL) seeks to overcome these challenges by exploring in parameters space that capture the correlation of actions. However, these approaches typically compromise data efficiency, as they treat trajectories as opaque \emph{black boxes}. In this work, we introduce a novel ERL algorithm, Temporally-Correlated Episodic RL (TCE), which effectively utilizes step information in episodic policy updates, opening the 'black box' in existing ERL methods while retaining the smooth and consistent exploration in parameter space. TCE synergistically combines the advantages of step-based and episodic RL, achieving comparable performance to recent ERL methods while maintaining data efficiency akin to state-of-the-art (SoTA) step-based RL.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2401.11437")
get_code_for_paper("2401.11437")
have("2401.11437")
Connect an agent — have() is free.