Skip to content
GCC AI Research

Search

Results for "MoE"

State-of-the-Art Arabic Language Modeling with Sparse MoE Fine-Tuning and Chain-of-Thought Distillation

arXiv ·

Arabic-DeepSeek-R1 is an application-driven, open-source Arabic Large Language Model (LLM) that has achieved a new state-of-the-art (SOTA) across the Open Arabic LLM Leaderboard (OALL). The model utilizes a sparse Mixture-of-Experts (MoE) backbone and a four-phase Chain-of-Thought (CoT) distillation scheme, which incorporates Arabic-specific linguistic verification and regional ethical norms. It records the highest average score on the OALL suite and outperforms proprietary frontier systems like GPT-5.1 on a majority of benchmarks evaluating comprehensive Arabic language-specific tasks. Why it matters: This work offers a validated and cost-effective framework for developing high-performing, culturally-grounded AI for under-represented languages, addressing the digital equity gap.

Ministry of Energy and Infrastructure Partners with Technology Innovation Institute to Build 3D Maps of UAE’s Mineral and Renewable Resources

TII ·

The UAE's Ministry of Energy and Infrastructure (MoEI) and the Technology Innovation Institute (TII) have partnered to create 3D maps of the UAE's mineral and renewable resources. TII's Directed Energy Research Center (DERC) will support MoEI in this effort, contributing its expertise to identify resources like geothermal energy and analyze geological data. The collaboration aims to support the UAE's Net Zero 2050 Strategy by enabling the exploration and utilization of undiscovered renewable and mineral resources. Why it matters: This initiative leverages local expertise to map domestic resources, aligning technological advancement with sustainability goals for the UAE.

Nile-Chat: Egyptian Language Models for Arabic and Latin Scripts

arXiv ·

The authors introduce Nile-Chat, a collection of LLMs (4B, 3x4B-A6B, and 12B) specifically for the Egyptian dialect, capable of understanding and generating text in both Arabic and Latin scripts. A novel language adaptation approach using the Branch-Train-MiX strategy is used to merge script-specialized experts into a single MoE model. Nile-Chat models outperform multilingual and Arabic LLMs like LLaMa, Jais, and ALLaM on newly introduced Egyptian benchmarks, with the 12B model achieving a 14.4% performance gain over Qwen2.5-14B-Instruct on Latin-script benchmarks; all resources are publicly available. Why it matters: This work addresses the overlooked aspect of adapting LLMs to dual-script languages, providing a methodology for creating more inclusive and representative language models in the Arabic-speaking world.

MoEI showcases National Data Center Observatory, signs MoU with 42 Abu Dhabi at Digital Readiness Retreat 2026 - MSN

The National ·

The Ministry of Energy and Infrastructure (MoEI) showcased its National Data Center Observatory at the Digital Readiness Retreat 2026. During the event, MoEI signed a Memorandum of Understanding (MoU) with 42 Abu Dhabi, a prominent coding school. This initiative aims to enhance the UAE's digital infrastructure and readiness capabilities. Why it matters: This development signifies the UAE government's commitment to strengthening its digital economy and fostering talent through strategic partnerships.

Quantum Research Center’s Dr. Giampiero Marchegiani Co-authors Research Paper on Bipolar Thermoelectricity

TII ·

Dr. Giampiero Marchegiani from the Quantum Research Center (QRC) co-authored a paper in Nature Nanotechnology on bipolar thermoelectricity in Josephson junctions. The research demonstrates the generation of electrical current using heat, observing bipolar thermoelectricity where the device generates either positive or negative electrical signals without extra control. Integrating these junctions into a bipolar thermoelectric Josephson engine (BTJE) allowed the generation of up to ~140nW/mm2 of electric power at subkelvin temperatures. Why it matters: This BTJE engine may have applications in superconducting technology, potentially strengthening the UAE's position in quantum technologies.

Partnership builds science acumen among young Arab readers

KAUST ·

KAUST and Frontiers for Young Minds (FYM) have partnered to produce child-friendly science articles in Arabic, engaging young Arabic speakers in scientific research. Their collaboration resulted in the first Arabic-original science article, "The Story of Legionella: From Surviving inside Amoebas to Causing Pneumonia", written by researchers from the University of Louisville. The article was reviewed by children, including two Saudis, with mentorship from a KAUST PhD graduate. Why it matters: This initiative promotes STEAM leadership in the region by making complex scientific concepts accessible to young Arabic-speaking readers, fostering early engagement in science.

Dow SISCA Awards celebrate KAUST and Dow’s joint commitment to sustainability

KAUST ·

The Dow SISCA awards at KAUST recognized two Ph.D. students for their sustainability-focused research. Anas Abu-Taha won first place and $10,000 for his work on nanoscale thermoelectrics for waste heat energy harvesting. David Conchouso Gonzales was the runner-up, receiving $2,500 for his poster on sustainable production of controllable emulsions using microfluidic droplet generators. Why it matters: The awards highlight KAUST's dedication to sustainability research and its ongoing partnership with Dow Chemical in fostering innovative solutions.

Understanding the mixture of the expert layer in Deep Learning

MBZUAI ·

A Mixture of Experts (MoE) layer is a sparsely activated deep learning layer. It uses a router network to direct each token to one of the experts. Yuanzhi Li, an assistant professor at CMU and affiliated faculty at MBZUAI, researches deep learning theory and NLP. Why it matters: This highlights MBZUAI's engagement with cutting-edge deep learning research, specifically in efficient model design.