Microsoft was founded on April 4, 1975. Who wants a piece of cake? 🎉

Visualization-of-Thought Elicits Spatial Reasoning in LLMs Inspired by a human cognitive capacity to imagine unseen worlds, this new work proposes Visualization-of-Thought (VoT) prompting to elicit spatial reasoning in LLMs. VoT enables LLMs to "visualize" their reasoning

Visualization-of-Thoughts (VoT): Mind's Eye of LLMs

Visualization-of-Thought Elicits Spatial Reasoning in Large Language Models VoT prompting enhances LLMs' spatial reasoning, enabling them to outperform MLLMs in spatial tasks. arxiv.org/abs/2404.03622

#Kosmos-2.5 available at aka.ms/kosmos25

$100 QMtech mister fpga board works perfectly for any devices including LCD and CRT displays MiSTer FPGA #MiSTerFPGA #Retrogaming

🚀 Exciting news! We’ve finally cracked the code for BitNet Hugging Face ! no pre-training needed! With just fine-tuning a Llama 3 8B, we've achieved great results, reaching a performance close to Llama 1 & 2 7B models on key downstream tasks! Want to learn more? Check out the

RedStone: Curating General, Code, Math, and QA Data for Large Language Models 🔗: github.com/microsoft/RedS… paper: arxiv.org/abs/2412.03398

RedStone: data pipeline designed to create specialized large-scale datasets by leveraging the vast and diverse data from Common Crawl. This paper from Microsoft introduce REDSTONE, an innovative and scalable pipeline engineered to extract and process data from Common Crawl,

In this issue of Research Focus, we examine a new conversation segmentation method that delivers more coherent and personalized agent conversation, and we review efforts to improve MLLMs’ understanding of geologic maps. Check out the latest research: msft.it/6019q9k33

Boosting consistency in generative games! Our "Model as a Game" paper introduces novel LogicNet & external map modules for numerical/spatial coherence w/ low overhead. See results in Traveler, Pong, Pac-Man. #GenAI #GameDev #MaaG Preprint: arxiv.org/abs/2503.21172

This is the first (small) large-scale training of native 1-bit LLMs / BitNet b1.58. More are coming soon including BitNet v2.

Geometric-Mean Policy Optimization Using geometric mean for the importance ratio, similar to GSPO (arxiv.org/abs/2507.18071).

New paper: #GMPO beats GRPO by simply switching from arithmetic → geometric mean for token rewards! ✅ More stable training (no extreme importance sampling ratios) ✅ Better exploration (higher entropy throughout training) huggingface.co/papers/2507.20…

RL(LLM) - Pisałem ostatnio o GSPO. A dzisiaj publikacje na temat -> GMPO - Geometric-Mean Policy Optimization, ARPO - Agentic Reinforced Policy Optimization, IRL - Inverse RL … Chyba najbardziej kwitnący obszar treningowy LLM. U nas Bielik-v3 też już trenowany RL (GRPO,

[CL] Geometric-Mean Policy Optimization Y Zhao, Y Liu, J Liu, J Chen... [Microsoft Research] (2025) arxiv.org/abs/2507.20673

8. Geometric-Mean Policy Optimization 🔑 Keywords: Geometric-Mean Policy Optimization, Policy Updates, Token-Level Rewards, Multimodal Reasoning, AI Native 💡 Category: Natural Language Processing 🌟 Research Objective: - The research aims to stabilize policy updates in

Microsoft Research introduces Geometric-Mean Policy Optimization (GMPO)! A new RL method that stabilizes LLM reasoning by maximizing the geometric mean of token-level rewards. No more unstable updates!

GMPO outperforms GRPO by 4.1% on math & 1.4% on multimodal reasoning benchmarks. It achieves better stability and performance, moving us closer to reliable AI. Learn more & get the code: Paper: huggingface.co/papers/2507.20… Code: github.com/callsys/GMPO

Top AI Papers of The Week (July 28 - August 3): - GEPA - Graph-R1 - AlphaEarth - Self-Evolving Agents - Hierarchical Reasoning Model - Efficient Attention Mechanisms - Geometric-Mean Policy Optimization Read on for more: