Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Flow to Control: Offline Reinforcement Learning with Lossless Primitive Discovery
Authors: Yiqin Yang, Hao Hu, Wenzhe Li, Siyuan Li, Jun Yang, Qianchuan Zhao, Chongjie Zhang
AAAI 2023 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The experimental results and extensive ablation studies on the standard D4RL benchmark show that our method has a good representation ability for policies and achieves superior performance in most tasks. |
| Researcher Affiliation | Academia | 1Department of Automation, Tsinghua University 2Institute for Interdisciplinary Information Sciences, Tsinghua University 3Harbin Institute of Technology |
| Pseudocode | Yes | Algorithm 1: IQL+LPD algorithm |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described. While it states 'We reproduce OPAL with authors providing code via email', this refers to a baseline, not their own method's code release. |
| Open Datasets | Yes | We evaluate our method on a suite of standard and challenging offline tasks (e.g., D4RL (Fu et al. 2020)) including Franka kitchen, Antmaze, and Adroit. |
| Dataset Splits | No | The paper mentions using D4RL tasks but does not specify how the data within these tasks are split into train/validation/test sets in the provided text. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details with version numbers (e.g., library or solver names with version numbers like Python 3.8, CPLEX 12.4). |
| Experiment Setup | Yes | Each experiment result is averaged over five random seeds with a standard deviation. We test the performance of IQL+OPAL with the most suitable steps c {1, 10} and fine-tune the expectile ratio λ and temperature parameter β in IQL. We ran IQL+LPD on kitchen-partial-v0 with various parameters, such as the expectile ratio λ [0.45, 0.9] and the temperature β [0.35, 0.8]. |