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].

Continuous-Time Reward Machines

Authors: Amin Falah, Shibashis Guha, Ashutosh Trivedi

IJCAI 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we evaluate the performance of our proposed approaches across benchmark environments to assess their efficiency and effectiveness. ... Figure 2 presents the performance of each approach across four benchmarks.
Researcher Affiliation Academia Amin Falah1 , Shibashis Guha2 , and Ashutosh Trivedi1 1University of Colorado Boulder 2Tata Institute of Fundamental Research EMAIL, EMAIL, EMAIL
Pseudocode No The paper describes algorithms conceptually and mathematically, such as the Q-value update equations (1) and (2) and the counterfactual experience generation process. However, it does not include a clearly labeled "Pseudocode" or "Algorithm" block with structured steps.
Open Source Code Yes Our implementation can be accessed at: https://github.com/falahamin1/Continuous-Time-Reward-Machines.git
Open Datasets No The paper describes custom 'benchmark environments' such as an 'autonomous vehicle in an urban environment' and a 'treasure hunt', which are inspired by existing works but no concrete access information (specific links, DOIs, or formal citations with authors/year) is provided for the datasets or environment definitions used in their experiments. For example, for the autonomous vehicle, it states 'inspired by [Oumaima et al., 2020]' but doesn't provide a link to the environment or data itself.
Dataset Splits No The paper describes a reinforcement learning setup where an agent interacts with an environment over 'episodes' and 'steps', rather than using pre-defined train/test/validation splits from a static dataset. It does not provide specific dataset split information.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory specifications) used for running the experiments.
Software Dependencies No The paper does not list any specific software dependencies or library versions (e.g., Python, PyTorch, TensorFlow, or other solvers with version numbers) used for the implementation.
Experiment Setup Yes For all experiments, we set the discount parameter α = 0.001 and learning rate θ = 0.1. The RL algorithms follow an ϵ-greedy exploration strategy, with ϵ initially set to 0.7 and decaying exponentially at a rate of 0.01 to transition from exploration to exploitation.