Search the archive

AMAZON LAUNCHES INTERESTS FOR PERSONALIZED SHOPPING

Amazon has introduced an AI-powered feature that tailors product discovery based on user-defined prompts that automatically surfaces relevant items and deals.

OPENAI ADOPTS RIVAL ANTHROPIC'S STANDARD FOR CONNECTING AI MODELS TO DATA

OpenAI will support Anthropic's Model Context Protocol (MCP) across its products, enhancing AI models' data-driven capabilities. MCP, an open-source standard, allows AI models to connect with various data sources like business tools and software. Companies like Block, Replit, and Sourcegraph have adopted MCP. OpenAI plans to share more about its integration soon.

ELON MUSK'S GROK AI LANDS ON TELEGRAM, GAINING ACCESS TO OVER 1 BILLION USERS: A NEW ERA FOR SEARCH AND CHAT

Elon Musk's AI chatbot, Grok, is now available on Telegram, offering its sarcastic assistant to over 1 billion users. Bundled with Telegram Premium, Grok expands beyond X, enhancing the platform with real-time reasoning and coding capabilities. While it raises privacy concerns, this move positions Telegram in the competitive AI landscape alongside giants like ChatGPT and Google's Gemini.

QWEN OMNI MODEL (HUGGING FACE HUB)

Qwen has released a model that can natively understand many different modalities while responding in either text or speech.

A MODULAR RAG FRAMEWORK

RGL streamlines the RAG pipeline for graph-structured data with modular components and optimized performance. It offers up to 143x speedup over traditional methods.

DETECTING SUBTLE DEEPFAKE MANIPULATIONS

Leveraging CLIP's visual encoder and parameter-efficient tuning, this method robustly detects partial deepfake forgeries across diverse datasets with minimal model changes.

here.

FRAME AUTOREGRESSIVE VIDEO MODELING

FAR improves video generation by modeling temporal dependencies between frames and introduces FlexRoPE to better handle long-range contexts using temporal decay.

ENHANCED MULTIMODAL VISION FOUNDATION MODELS

The CoMP pipeline improves cross-modal alignment and resolution adaptability in vision-language models through continual pretraining and custom embeddings.

IMPROVED 3D GAUSSIAN SPLATTING WITH HOGS

HoGS boosts rendering accuracy of distant objects in 3DGS pipelines using homogeneous coordinates, enabling better performance in outdoor scene reconstruction.

OPENAI RESEARCH LEAD NOAM BROWN THINKS CERTAIN AI 'REASONING' MODELS COULD'VE ARRIVED DECADES AGO

AI reasoning models could have emerged 20 years earlier with the right approach and algorithms. Academic institutions can still contribute by focusing on areas requiring less computing, such as model architecture design. The poor state of AI benchmarks is an area where academia can make a significant impact.

AI COMPANIONS: WEIRD AND CREEPY OR WORTH YOUR ATTENTION?

AI companions, which offer personalized and empathetic interactions, are rapidly transitioning from fantasy to reality. They are particularly appealing to Gen Z and Gen Alpha for safe exploration and engagement. Despite challenges like quirky interactions and emotional boundary concerns, AI companions provide substantial value by enhancing online experiences, storytelling, and personalized learning. Companies like Genies are leveraging avatar technology and generative AI to create immersive, context-aware digital companions that aim to form authentic connections with users.

HOW TRUMP'S AI CZAR PLANS TO REWRITE US TECH POLICY

President Trump has appointed David Sacks as AI and cryptocurrency czar. Tasked with leading the Presidential Council of Advisors for Science and Technology, Sacks, alongside others, is charged with crafting a new AI action plan to replace what he describes as burdensome regulations from the previous administration. Sacks has raised concerns about China's AI advancements and emphasized the need for the US to ease regulations to maintain its leadership in AI.

APPLE FACES LAWSUIT OVER APPLE INTELLIGENCE DELAYS

Apple is facing a class-action lawsuit for allegedly misleading advertising regarding its Apple Intelligence features.

HOLLYWOOD VS AI: 400 CELEBS URGE TRUMP TO REIN IN OPENAI, GOOGLE

Hollywood creatives have signed an open letter urging the U.S. government not to weaken copyright protections for AI companies like OpenAI and Google.

AI-DRIVEN WEATHER PREDICTION BREAKTHROUGH REPORTED

Aardvark Weather introduces an AI approach capable of delivering faster and more accurate weather forecasts using significantly less computing power than traditional systems.

Read the OpenAI playbook

OPENAI'S IMPROVED IMAGE GENERATION

OpenAI's GPT-4o introduces improved image generation with precise text rendering, instruction following, and multi-turn editing.

DEEPSEEK-V3-0324 RELEASE WITH MIT LICENSE

DeepSeek has released its new V3-0324 model, which outperforms GPT 4.5 in most benchmarks and features major improvements in performance.

QWEN 2.5 32B VISION LANGUAGE MODEL

Qwen has released a strong vision language model that is open and can run reasonably well on consumer hardware.

MODIFYING LARGE LANGUAGE MODEL POST-TRAINING FOR DIVERSE CREATIVE WRITING

Midjourney has released some work to improve diversity in creative writing models. It was able to post-train a small 7B model that outperforms much larger open and closed models in creative writing.

VIDEO T1

Test time compute for video that uses a guidance model to reject frame paths that don't conform to physics or the user specified prompt. Using test time compute for this task dramatically improves performance on benchmarks.

3D FACE EDITING

FFaceNeRF enhances 3D face editing by overcoming fixed mask limitations in NeRF-based methods.

VISUAL GEOMETRY GROUNDED TRANSFORMER

VGGT is a feed-forward neural network that directly infers all key 3D attributes of a scene, including extrinsic and intrinsic camera parameters, point maps, depth maps, and 3D point tracks, from one, a few, or hundreds of its views, within seconds.

DEREFLECTION ANY IMAGE

Dereflection Any Image (DAI) introduces a new diffusion-based approach for reflection removal, leveraging a high-quality dataset and progressive training.

SLOW-THINKING REASONING MODEL WITH FASTCURL

FastCuRL-1.5B-Preview advances slow-thinking reasoning models with curriculum reinforcement learning, achieving state-of-the-art results in fewer training steps.

SINGLE IMAGE ITERATIVE SUBJECT-DRIVEN GENERATION AND EDITING

SISO is an inference-time optimization method that personalizes images from a single subject image without training. It can personalize the subject of a given image or generate new images with the personal subject.

MANNY MEDINA'S NEW STARTUP: PAID

Outreach founder Manny Medina has launched Paid, a platform designed to help AI agent startups manage pricing and profitability. With backing from top investors, Paid aims to redefine how AI agents are billed and valued in the emerging agent economy.

A KEY DEEPMIND ROBOTICS RESEARCHER LEFT GOOGLE, AND NVIDIA HAS ALREADY BACKED HIS STEALTH STARTUP

Pete Florence, previously a senior research scientist at DeepMind, has left to form Generalist AI, a robotics startup backed by Nvidia's VC arm, NVentures. The company aims to create general-purpose robots, with Florence highlighting the vision of reducing the marginal cost of physical labor to zero. Florence joins other former DeepMind scientists in launching tech ventures amidst Nvidia's growing influence in AI.

CLAUDE CAN NOW SEARCH THE WEB

Claude now features web search functionality for real-time insights and up-to-date responses, complete with source citations.

INTRODUCING NEXT-GENERATION AUDIO MODELS IN THE API

OpenAI has introduced new audio models, including Voice Engine, capable of generating realistic voices from short samples.

PERPLEXITY IS REPORTEDLY IN TALKS TO RAISE UP TO $1B AT AN $18B VALUATION

AI search startup Perplexity is in talks to raise $1 billion at a valuation of $18 billion.

If you're building an AI business, achieving ISO 42001 compliance

shows your customers that you are taking the necessary steps to ensure responsible usage and development of AI. But this can cost you real time and  money.

INTRODUCING TOGETHER CHAT: USE DEEPSEEK R1 FOR FREE, HOSTED IN NORTH AMERICA

Together Chat has launched a free consumer app that offers seamless interaction with top open-source models, including the new DeepSeek R1 for advanced reasoning. It supports web crawling, code generation with Qwen Coder 32B, and image creation using Flux Schnell, all accessible via a Progressive Web App.

ROBLOX'S NEW AI MODEL CAN GENERATE 3D OBJECTS

Roblox has open-sourced Cube 3D, an AI model that can generate 3D objects from text prompts, to enhance creation efficiency. Cube 3D uses tokenization techniques and is trained with licensed and publicly available datasets, plus Roblox experience data. Future iterations will incorporate multimodal inputs like images and videos.

NEW REVE IMAGE GENERATOR BEATS AI ART HEAVYWEIGHTS MIDJOURNEY AND FLUX AT A PENNY PER IMAGE

Reve Image 1.0, an AI image generator, offers competitive pricing and impressive prompt adherence, realism, and versatility, potentially outperforming rivals like Midjourney and Ideogram. At $5 for 500 credits, it allows affordable generation of high-quality images but lacks an edit feature and mobile app, which may deter advanced users. Despite limited transparency about its team and technology, Reve remains a strong cost-effective choice for users seeking high-quality outputs without extensive technical know-how.

LLAVA-MORE

LLaVA-MORE systematically evaluates different language models and visual backbones in multimodal LLMs, providing a reproducible framework for comparing architectures.

SKELETON-BASED ACTION SEGMENTATION

TRG-Net enhances action segmentation using text-derived relational graphs for improved spatial-temporal modeling and supervision.

HIERARCHICAL DOCUMENT STRUCTURE ANALYSIS

UniHDSA is a unified relation prediction approach for hierarchical document structure analysis, enabling a single module to handle multiple tasks simultaneously.

SINGLE IMAGE PERSONALIZATION WITHOUT TRAINING

SISO enables training-free personalization for image generation and editing by optimizing similarity loss iteratively.

AWESOME MRI RECONSTRUCTION

A curated collection of articles about deep learning in MRI reconstruction.

OPEN-SOURCE OMNI-MODAL FOUNDATION MODEL

The Baichuan Omni 1.5 model supports text, image, video, and audio inputs as well as text and audio outputs. This means it is another example of an any-to-any style model, sometimes referred to as natively multimodal. The model, like many, uses an interleaved multimodal tokens approach where different types of tokens are routed to different encoders/decoders that all get processed by some main auto-regressive model.

MCP (MODEL CONTEXT PROTOCOL): SIMPLY EXPLAINED IN 5 MINUTES

MCP (Model Context Protocol) lets AI tools like Claude and ChatGPT integrate with your everyday apps—turning them from isolated chatbots into powerful assistants with real-world capabilities. Instead of copying error logs into your AI, MCP lets you simply say "read my browser console and fix this bug" and the AI can directly access those tools. Behind the scenes, MCP works through a standard interface where providers like Slack, GitHub, and Sentry can create "adapters" that translate AI requests into API calls.

FINE-TUNE GEMMA 3 FOR FREE

The Unsloth team has figured out some of the quirks with the new open weights model from DeepMind. By integrating with their toolkit, one can train the model on a free Colab instance.

OPENAI'S O1-PRO IS THE COMPANY'S MOST EXPENSIVE AI MODEL YET

OpenAI's o1-pro, a more computing-intensive version of its o1 reasoning AI, is now available in its developer API for select users at a premium cost. Priced at $150 per million input tokens and $600 per million output tokens, it offers better performance but has received mixed early reviews as it struggles with certain tasks. Despite improvements, internal benchmarks show only slight performance gains over the standard o1 model in coding and math.

linear models to neural networks; and more. Listen here