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..

MaNGO — Adaptable Graph Network Simulators via Meta-Learning

Authors: Philipp Dahlinger, Tai Hoang, Denis Blessing, Niklas Freymuth, Gerhard Neumann

NeurIPS 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We validate our approach, Meta Neural Graph Operator (Ma NGO), on several dynamics prediction tasks with varying material properties, demonstrating superior performance over existing GNS methods.
Researcher Affiliation Academia Philipp Dahlinger Tai Hoang Denis Blessing Niklas Freymuth Gerhard Neumann Autonomous Learning Robots Karlsruhe Institute of Technology Karlsruhe
Pseudocode No The paper describes the methodology in prose and mathematical equations, accompanied by architectural diagrams (e.g., Figure 3), but does not include any explicitly labeled pseudocode or algorithm blocks.
Open Source Code Yes Code: https://github.com/ALRhub/mango Dataset: https://zenodo.org/records/17287535
Open Datasets Yes Code: https://github.com/ALRhub/mango Dataset: https://zenodo.org/records/17287535
Dataset Splits Yes Table 1: Dataset descriptions Name Train/Val/Test Splits Number of Steps Number of Nodes for Prediction PB 460/50/50 50 225 DP-easy 600/100/100 52 81 DP-hard 600/100/100 52 81 SCC 600/100/100 100 400 (cloth) + 98 (sphere)
Hardware Specification Yes The training took place on an NVIDIA A100 GPU, with each method given the same computation budget of 48 hours.
Software Dependencies Yes This dataset is created using NVIDIA Isaac Sim [65], which leverages Phys X 5.0 [66] to simulate position-based dynamics (PBD) particle interactions.
Experiment Setup Yes Table 3: Left: Training setup for each dataset. Right: Noise-scale per task for Auto-regressive methods