Latent Variable Representation for Reinforcement Learning

Authors: Tongzheng Ren, Chenjun Xiao, Tianjun Zhang, Na Li, Zhaoran Wang, sujay sanghavi, Dale Schuurmans, Bo Dai

ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirically, we demonstrate superior performance over current state-of-the-art algorithms across various benchmarks. We extensively test our algorithm on the Mujoco (Todorov et al., 2012) and Deep Mind Control Suite (Tassa et al., 2018).
Researcher Affiliation Collaboration 1Google Research, Brain Team 2UT Austin 3University of Alberta 4UC Berkeley 5Harvard University 6Northwestern University 7Georgia Tech
Pseudocode Yes Algorithm 1 Online Exploration with LV-Rep
Open Source Code Yes Project Website: https://rlrep.github.io/lvrep/
Open Datasets Yes We extensively test our algorithm on the Mujoco (Todorov et al., 2012) and Deep Mind Control Suite (Tassa et al., 2018).
Dataset Splits No The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper mentions 'Pytorch' and an 'open source implementation' for SAC baseline with a citation but does not provide specific version numbers for these software components.
Experiment Setup Yes Table 3: Hyperparameters used for LV-Rep in all the environments in Mu Jo Co and DM Control Suite. Hyperparameter Value Actor lr 0.0003 Model lr 0.0003 Actor Network Size (Mu Jo Co) (256, 256) Actor Network Size (DM Control) (1024, 1024) LV-Rep Feature Embedding Dim (Mu Jo Co) 256 LV-Rep Feature Embedding Dim (DM Control) 1024 ERP Embedding Network Size (DM Control) (1024, 1024, 1024) Critic Network Size (Mu Jo Co) (256, 256, 1) Critic Network Size (DM Control) (1024, 1024, 1) Discount 0.99 Critic Target Update Tau 0.005 Latent Variable Target Update Tau 0.005 Batch Size 256