ContraBAR: Contrastive Bayes-Adaptive Deep RL

Authors: Era Choshen, Aviv Tamar

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

Reproducibility Variable Result LLM Response
Research Type Experimental Our third contribution is an empirical evaluation of our method that exposes several advantages of the contrastive learning approach.
Researcher Affiliation Academia 1Technion, Haifa, Israel. Correspondence to: Era Choshen <erachoshen@campus.technion.ac.il>.
Pseudocode No The paper describes the Contra BAR algorithm and its architecture using text and diagrams (e.g., Figure 1) but does not provide a formal pseudocode block or algorithm listing.
Open Source Code Yes exact hyperparameters can be found in our code at https://github.com/ec2604/Contra BAR
Open Datasets No The paper describes using specific environments for training, such as MuJoCo locomotion tasks and custom image-based environments (Reacher-Image, Panda Reacher, Panda Wind). While these are based on existing simulators, the paper does not provide concrete access (link, DOI, citation with author/year) to the specific datasets collected or used for training these experiments.
Dataset Splits No The paper discusses 'training tasks' and evaluating 'test performance' but does not explicitly provide details about training/validation/test dataset splits, such as specific percentages, sample counts, or references to predefined splits.
Hardware Specification No The paper does not explicitly describe the specific hardware used for running its experiments, such as GPU models, CPU types, or memory details. It only mentions general computational aspects like 'memory limitations'.
Software Dependencies No The paper mentions using algorithms like PPO (Schulman et al., 2017) and SAC (Haarnoja et al., 2018), but it does not specify version numbers for any software dependencies, libraries, or frameworks (e.g., PyTorch, TensorFlow, Python version).
Experiment Setup Yes The paper provides specific details about the experimental setup within the environment descriptions, such as: 'The horizon is set to 50', 'radius r = 0.05', 'action space is 3-dimensional and bounded [ -1, 1]3', 'images of size 64 x 64'. It also explicitly states: 'exact hyperparameters can be found in our code at https://github.com/ec2604/Contra BAR'.