Search the archive

GAUSSIAN TRANSFORMER FOR 3D SPATIAL UNDERSTANDING

This paper introduces GaussTR, a novel Gaussian Transformer that leverages alignment with foundation models to advance self-supervised 3D spatial understanding.

3D CAD REVERSE ENGINEERING

An open-source tool for Computer-Aided Diagnosis, offering a modular and scalable platform for medical imaging research and development.

AI IS BOOMING ON THE APP STORE, AND DEVELOPERS ARE TAKING ADVANTAGE OF IT

AI apps are flooding the App Store charts, but many prioritize cash grabs over quality, often misleading users with paywalls and subpar features. HUBX, dominating with apps like DaVinci AI, faces criticism for poor functionality and misleading marketing despite high rankings. Amidst the saturation, quality apps like Photoroom and Picsart AI stand out by delivering on their promises with effective AI tools.

2024 BACKWARD PASS: THE DEFINITIVE GUIDE TO AI IN 2024

Kelvin My from Translink Capital shares a recap of everything that's happened in AI in 2024. The report provides an overview of each of AI's four layers: infrastructure, foundational model, tooling, and application. For each layer, it discusses key takeaways, future trends in 2025 and beyond, and lists some noteworthy startups.

THINKING IN SPACE

A research initiative focused on spatial reasoning and AI models designed to interpret and interact within three-dimensional spaces.

SERVERLESS LORA

Together AI has a new product that enables users to deploy a custom LoRA at the cost of the base model via serverless switching.

BLOOD TESTS ARE FAR FROM PERFECT — BUT MACHINE LEARNING COULD CHANGE THAT

Researchers at the University of Washington and Harvard have used machine learning to create personalized blood test references, enhancing disease prediction accuracy.

OPENAI COFOUNDER ILYA SUTSKEVER SAYS THE WAY AI IS BUILT IS ABOUT TO CHANGE

Ilya Sutskever predicts that pre-training in AI will end due to the limited availability of new data.

👉Save your spot

1-800-CHATGPT

For day 10, OpenAI has released free voice mode and texting via WhatsApp for a limited number of minutes per month. It operates globally and uses the Advanced Voice Mode API.

GITHUB COPILOT NOW FREE FOR VS CODE

GitHub has made Copilot free to use within Visual Studio Code, democratizing AI-powered coding assistance for developers.

THE SECOND GEMINI

Google has released Gemini Flash 2.0, which features capabilities like deep research, a real-time multimodal API, and a functional code interpreter. Alongside, projects like Astra, Mariner, and Jules are in experimental stages, focusing on universal AI assistance, web reasoning, and code automation, respectively. While Google introduces these innovations, clear communication on capabilities remains an area for improvement.

ALIGNMENT FAKING IN LARGE LANGUAGE MODELS

Interesting research from Anthropic and Redwood that explores what models do if they know they are being aligned. Interestingly, they show alignment while still maintaining their original preferences. This is helpful to show holes in existing alignment methods.

ARE YOUR LLMS CAPABLE OF STABLE REASONING?

Reasoning is a huge opportunity for models, especially for critical applications like real world deployment. However, benchmarks that are meant to test reasoning ability don't show how stable a model is across novel tasks. This paper introduces a new benchmark called G-Pass@k which evaluates a model's peak and stable reasoning ability.

FACTORIZING TRANSFORMERS FOR LINEAR TIME INFERENCE

This paper introduces a new transformer design that achieves linear time complexity during inference by factorizing the attention mechanism, reducing computational costs while maintaining model performance.

BRUSHEDIT

All in one model and system for image inpainting and editing. It breaks the task up into a number of sequences for editing, masking, inpainting, and so on. It uses pre-trained VLLMs (like GPT-4o) to improve object understanding and masking.

SELECTIVE TOKEN REMOVAL

A tool designed for selectively erasing tokens from text while preserving context, aimed at improving text anonymization workflows.

OPEN-SOURCE VIDEO TOKENIZER

VidTok is a versatile video tokenizer that delivers state-of-the-art performance in both continuous and discrete tokenizations.

INTRODUCTION TO GENIES' SMART AVATARS

Genies introduces Smart Avatars, AI-driven digital beings that revolutionize online interactions by serving as dynamic identity extensions. Leveraging LLMs and behavioral AI, these avatars enhance digital experiences across games and platforms and offer opportunities for monetization and user engagement.

PERPLEXITY'S CAMPUS STRATEGIST PROGRAM

Perplexity AI announces its 2024 program to engage students in AI adoption, offering campus-exclusive resources and collaborations.

AETHIR AND PARTNERS POUR $40M INTO DECENTRALIZED INFRASTRUCTURE FOR AI AND BLOCKCHAIN

Aethir, in collaboration with Beam Foundation, Sophon Foundation, and Permian Labs, is launching Tactical Compute (TACOM) with a $40 million initiative to provide decentralized GPU infrastructure. TACOM aims to meet the rising demand for scalable compute power for AI, gaming, and blockchain through tokenized and distributed solutions. The project will enable new opportunities for GPU resource monetization and drive innovation in AI and decentralized ecosystems.

lets you write, and reproducibility in training runs. Listen

META LAUNCHES OPEN SOURCE LLAMA 3.3, SHRINKING POWERFUL BIGGER MODEL INTO SMALLER SIZE

Meta's Llama 3.3 is a cost-efficient open-source LLM with 70 billion parameters that offers performance on par with larger models like the 405B Llama 3.1, but with significantly reduced GPU and power costs.

MICROSOFT UNVEILS ZERO-WATER DATA CENTERS TO REDUCE AI CLIMATE IMPACT

Microsoft is introducing a zero-water cooling system for its data centers to eliminate the annual use of over 125 million liters of water per center.

SURREY ANNOUNCES WORLD'S FIRST AI MODEL FOR NEAR-INSTANT IMAGE CREATION ON CONSUMER-GRADE HARDWARE

The Surrey Institute for People-Centred AI focuses on creating ethical, responsible, and inclusive AI with a people-first approach.

Delete your personal data today.

🎁 HOLIDAY SPECIAL: Use this link [4] to get 58% off using code TLDRAI.

NVIDIA'S GENAI SUPERCOMPUTER

NVIDIA has unveiled its most affordable generative AI supercomputer, “Jetson Orin Nano Super Developer Kit”.

OPENAI'S DEVELOPER APIS

OpenAI introduces demo developers and updates APIs.

ARM CEO RENE HAAS ON THE AI CHIP RACE, INTEL, AND WHAT TRUMP MEANS FOR TECH

Arm Holdings CEO Rene Haas discusses the rapid evolution of AI and its impact on the chip industry, emphasizing Arm's strategic role in providing architectures for a wide range of applications. He addresses Arm's potential move into designing AI chips, Intel's challenges, and the importance of balancing innovation with geopolitical considerations, especially in China. Haas also highlights Arm's growing presence in AI workloads across various devices, stressing the need for more compute capabilities to support AI's expansion.

A LONG-CONTEXT OPEN-WEIGHT LANGUAGE MODEL

Giraffe introduces a transformer-based architecture capable of processing significantly longer input contexts, pushing the boundaries of context length in open-weight models.

META'S ADAPTIVE CACHING FOR FAST VIDEO GENERATION

40+ YEARS OF SATELLITE DATA FOR ML RESEARCH

The Digital Typhoon Dataset is the longest satellite image dataset for typhoons, spanning over 40 years.

Register for free ->

RETROLLM

RetroLLM is a framework designed for training large language models with retrieval augmentation, enabling more efficient and accurate responses by integrating external knowledge sources during inference.

ITERATION OF THOUGHT FOR BETTER AI RESPONSES

Iteration of Thought (IoT) uses dynamic, thought-provoking prompts to improve LLMs. Unlike existing methods like Chain of Thought (CoT), IoT adapts to the context of each interaction.

A COST-EFFECTIVE ARCHITECTURE WITH TOKENFORMER

TokenFormer is a novel architecture designed to reduce the high computational costs associated with scaling transformer models.

GROK FOR EVERYONE

Grok has a new version and a new efficient model that is available for all users. It also has an improved image generation model and API.

AI WORLD MODELS

World models enable AI systems to simulate and reason about their environments, pushing forward autonomous decision-making and real-world problem-solving.

15 TIMES TO USE AI, AND 5 NOT TO

AI is useful for tasks like idea generation, summarization, and translation where quantity or varied perspectives are needed. It excels when humans can quickly assess AI's output and in tasks where minimal downside exists for errors. AI use can hinder learning or accuracy in high-stakes or unfamiliar situations, requiring careful judgment to balance its utility.

GROK IMAGE GENERATION RELEASE

Grok's new image generation model excels in rendering real-world details, logos, and realistic human portraits.

YOUTUBE'S NEW AUTO-DUBBING FEATURE IS NOW AVAILABLE FOR KNOWLEDGE-FOCUSED CONTENT

YouTube's auto-dubbing feature is now available to hundreds of thousands more channels, focusing initially on informational content.

GOOGLE KICKS OFF $20B RENEWABLE ENERGY BUILDING SPREE TO POWER AI

Google signed a $20 billion deal with Intersect Power and TPG Rise Climate to build renewable energy and storage for its data centers.

VEO 2 AND IMAGEN 3

Google has announced a new video model and a new image generation model. Both are stunning improvements over the previous iterations.

OPENAI SEARCH

OpenAI explores the potential of ChatGPT Search on the 8th day of its announcements.