โšก Open Science Initiative

Advancing Artificial General Intelligence

Our AGI Research Division focuses on scalable reasoning, value alignment, multimodal cognition, and open benchmarks. We publish transparently, collaborate globally, and release all datasets and code under permissive licenses.

47Published Papers
12Open Datasets
34Active Researchers
8.2kGithub Stars

Core Pillars

Our AGI program is structured around four foundational tracks, each addressing critical bottlenecks in path-to-general intelligence systems.

๐Ÿง 

System 2 Reasoning

Architectures that enable slow, deliberate, step-wise reasoning with self-correction, uncertainty quantification, and causal inference.

Active
โš–๏ธ

Value Alignment & Safety

Robust preference learning, constitutional AI frameworks, and red-teaming protocols to ensure corrigibility and human-compatible objectives.

Active
๐ŸŒ

Multimodal World Models

Unified representations across text, vision, audio, and embodied simulation for grounded, transferable understanding.

Active
๐Ÿ“Š

Open Benchmarks

Reproducible, contamination-resistant evaluation suites measuring reasoning, creativity, and cross-domain transfer without leaderboard gaming.

In Progress

Recent Papers

Peer-reviewed and pre-print research from the Aevum AGI Division. All code and supplementary materials are open-source.

arXiv:2411.0892Nov 2024

Chain-of-Thought Abstraction via Dynamic Pruning in LLM Reasoning

Dr. E. Vasquez, L. Chen, A. Mwangi (Aevum AGI)

We introduce a dynamic pruning mechanism that selectively collapses redundant reasoning steps while preserving logical validity, reducing inference latency by 42% without accuracy degradation on GSM8K and MMLU-Pro.

PDF Code
ICLR 2025Feb 2025

Constitutional Reward Modeling for Cross-Cultural Value Alignment

Prof. R. Al-Fayed, S. Johansson, K. Tanaka (Aevum AGI)

A novel reward modeling framework that integrates culturally-diverse preference datasets with constitutional constraints, demonstrating improved robustness against adversarial prompt injection and value misalignment.

PDF Code
NeurIPS 2024Dec 2024

Grounded Multimodal Reasoning through Embodied Simulation Priors

Dr. M. Okoro, J. Park, Aevum AGI Lab

We show that injecting physics-informed simulation priors into multimodal transformers significantly improves spatial reasoning and causal prediction in complex visual-language tasks.

PDF Code
Aevum Technical ReportJan 2025

Aevum-AGI-Bench: Contamination-Resistant Evaluation for General Reasoning

Aevum AGI Evaluation Team

A rolling benchmark suite with synthetic task generation, strict train/test separation protocols, and adversarial contamination detection to measure true generalization capabilities.

PDF Dataset

Datasets & Benchmarks

Freely available for academic and commercial research. All releases include documentation, citation guidelines, and usage licenses.

Dataset / Benchmark Domain Size / Scope License Access
Aevum-Reason-1M ReasoningLogic 1.2M samples CC-BY-4.0 Download
CrossCult-Align-50K AlignmentPreferences 50K pairs, 18 regions CC-BY-NC-4.0 Download
MultiSim-Embodied MultimodalPhysics 850K trajectory clips Apache-2.0 Download
Aevum-AGI-Bench v2.1 EvaluationRolling 12K dynamic tasks MIT Leaderboard

Research Team

Interdisciplinary scholars, ML engineers, and cognitive scientists driving our AGI initiative forward.

EV

Dr. Elena Vasquez

Director of AGI Research

RA

Prof. Rami Al-Fayed

Lead, Alignment & Safety

MO

Dr. Michael Okoro

Principal Scientist, Multimodal

LJ

Dr. Linna Johansson

Lead Engineer, Reasoning Architectures

Stay Updated on Our Research

Receive monthly research digests, dataset releases, and early access to pre-prints. We share our methodology, failures, and breakthroughs transparently.

For collaboration inquiries, contact: agi-research@aevum.edu