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 [1].

Guaranteed Discovery of Control-Endogenous Latent States with Multi-Step Inverse Models

Authors: Alex Lamb, Riashat Islam, Yonathan Efroni, Aniket Rajiv Didolkar, Dipendra Misra, Dylan J Foster, Lekan P Molu, Rajan Chari, Akshay Krishnamurthy, John Langford

TMLR 2023 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the discovery of the control-endogenous latent state in three domains: localizing a robot arm with distractions (e.g., changing lighting conditions and background), exploring a maze alongside other agents, and navigating in the Matterport house simulator.
Researcher Affiliation -1 Anonymous authors Paper under double-blind review. The paper does not provide clear institutional affiliations or email domains for the authors, as it is under double-blind review.
Pseudocode Yes Algorithm 1 AC-State Algorithm for Latent State Discovery Using A Uniform Random Policy
Open Source Code No The text does not contain an explicit statement by the authors releasing their code, nor does it provide a direct link to a code repository for the methodology described in this paper. While a third-party library 'vector-quantize-pytorch' is referenced, this is not the authors' own implementation code.
Open Datasets Yes We evaluated AC-State on the matterport simulator introduced in Chang et al. (2017).
Dataset Splits No The paper mentions collecting "14,000 samples" for the robot arm, "3,000 training samples" for the maze, and a "20,000 sample dataset" for Matterport, but does not specify any explicit training, validation, or test splits for these datasets.
Hardware Specification No The paper does not provide specific details about the hardware used for running its experiments, such as exact GPU/CPU models, processor types, or memory amounts.
Software Dependencies No The paper mentions various models and optimizers, such as MLP-Mixer (Tolstikhin et al., 2021), Adam optimizer (Diederik et al., 2014), and VQ-VAE (van den Oord et al., 2017), but does not provide specific version numbers for software libraries or programming environments like Python, PyTorch, or CUDA.
Experiment Setup Yes The model is trained end-to-end with the AC-State objective with maximum horizon of K = 5 for 20 epochs using the Adam optimizer with a learning rate of 1e-4. We use a 6-layer transformer with 256 dimensions in the embedding. We set the FFN dimension D to 512. We use 4 heads in the ViT.