Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
The "1.5-Pints" Language Model presents a new approach to compute-efficient pre-training. By curating a high-quality dataset of 57 billion tokens, this model surpasses Apple's OpenELM and Microsoft's Phi in instruction-following tasks, as measured by MT-Bench.
Lighthouse is a user-friendly library designed to tackle the challenges in video moment retrieval and highlight detection.
India's technology outsourcing industry is facing a pivot as AI integration threatens jobs, especially in basic tasks like call centers. Companies are challenged to ascend the value chain amidst a hiring slowdown, with some hopeful AI tools will create new opportunities in sectors like programming. The industry's future demands higher-level cognitive skills as automation continues to redefine traditional roles.
Breakthroughs in AI pre-training have enabled models to process massive online datasets, while subsequent supervised fine-tuning with experts helps models generalize and gain specialized knowledge. Turing's approach of using "input and output pairs" generated by subject-matter experts aims to advance AI reasoning capabilities. Anticipating an "agentic" era of AI, these models could autonomously perform complex tasks by integrating specialized knowledge across domains.
Perceptive has developed an AI-controlled autonomous dental robot that performed a complete dental procedure notably faster than traditional methods.
Microsoft's new "GitHub Models" offers developers access to AI models like Llama 3.1 and GPT-4 and enables AI integration within software projects.
Google is deploying new Gemini-powered features to Chrome Desktop, including Google Lens integration for visual searches, Tab Compare for streamlined shopping comparisons, and natural language processing for enhanced browsing history search.
Google announced new Gemini AI-powered features for Google Home, including intelligent captions for Nest camera footage, natural language processing for Home routine creation, and an upgraded, more natural-sounding Google Assistant. The advanced features, primarily behind a Nest Aware subscription paywall, aim to enhance the smart home experience with the rollout commencing in beta and expanding next year. As part of the move toward smarter home automation, Google envisions an assistant that can proactively manage complex and dynamic home environments.
Nvidia is facing increased government scrutiny from the EU, UK, China, and the US Justice Department over its dominant market share in AI chips and sales practices. The company is rapidly building its legal and policy teams to address antitrust concerns amid profitable growth, as it commands 90 percent of the GPU market essential for AI systems. Nvidia is also adapting to increased competition oversight, with recent attention turning to its planned acquisition of Run.ai and impact on the AI supply chain.
Meta plans to significantly increase computing capacity for training its next-gen large language model, Llama 4, expecting a 10x compute increase over Llama 3. The investment in AI training infrastructure will drive up capital expenditures in 2025. Despite heavy spending, Meta doesn't foresee immediate significant revenue from Gen AI products.
There is strong pressure to use compute at inference time to improve model performance. This paper showcases several methods that can be used and discusses the trade offs made between them. In general, this points towards a broader trend of squeezing performance out of smaller models.
There has been a push in vision language models to make them easy to tune and transfer between tasks. This is reminiscent of computer vision a decade ago. Importantly, this can be done through careful data curation and architecture updates which are on display with LLaVA-OneVision.
Treating 3D objects as UV wrapped images makes it simple to use a DiT model to synthesize novel objects based on textual inputs.
muP is a way to tune your hyperparameters to transfer from smaller to larger models. This GitHub gist shows a great theorem in practice which states that you can change where you scale model outputs and it doesn't affect final transfer performance.
XLabs has released the first Flux-Dev control net which allows for generation conditioned on Canny image inputs.
HARMONIC is a framework for generating and evaluating synthetic tabular data using large language models.
GitHub is leveraging AI and machine learning to analyze and summarize customer feedback at scale, informing product improvements and feature prioritization with actionable insights. This automated approach enables the capture of multilingual feedback and drives data-driven decision-making, enhancing the responsiveness to developer needs. The initiative highlights GitHub's commitment to integrating AI to maintain a developer-centric product development process.
This article presents a deep skepticism about OpenAI's sustainability, considering generative AI's lack of broad market utility and the prohibitive costs of developing and running large language models. The author questions OpenAI's long-term viability without significant technological breakthroughs or continuous, unprecedented fundraising efforts. Despite OpenAI's influence on the AI industry, the company faces challenges with profitability, high operational burn rates, and dependence on strategic partnerships, particularly with Microsoft.
AMD's Q2 2024 earnings show a significant shift toward data center products, with nearly half of its sales now in this area, thanks primarily to the Instinct MI300 AI chip. The company has committed to annual new AI chip releases, competing with Nvidia's offerings, despite supply constraints projected to last until 2025. Although Nvidia remains ahead in the data center market, AMD has seen growth in its CPU and GPU segments, including its Ryzen processors and Radeon 6000 GPUs.
Sponsored
For human-like AI voice generation, models need to be trained on diverse pronunciation data. Discover more with this on-demand webinar
Microsoft's AI investments surged to $19 billion last quarter, focusing on data centers for AI capabilities, but significant AI revenue remains uncertain.
Yum! Brands is expanding its AI-driven Voice AI technology to hundreds of Taco Bell drive-thrus to enhance efficiency, aiming for broader rollout by year-end.
OpenAI is leading a $60 million funding round for Opal, the same company behind the high-end Tadpole webcam.
Sponsored
that enables the precise segmentation of any object in any video or image.
Figure has released its newest humanoid robotics with new smarts and improved design.
Requesting JSON or other structured output from language models is challenging. This new feature in OpenAI's API supports structured output from language model generation that can be consumed by downstream deterministic programs.
Meta is negotiating with celebrities like Awkwafina and Judi Dench to use their voices for its AI digital assistant, seeking extensive usage rights across its platforms. Striking a deal could mean millions in fees for the actors, with safeguards from SAG-AFTRA for AI-generated likenesses. The company aims to present these AI features at its Connect conference in September, after recently canceling a celebrity voice chatbot project.
This work generates artistic quality meshes from 3D, 2D, or textual input. It uses improvements in the vertex representation and uses adjacent tokens to efficiently sample.
Researchers have developed an automated review generation method using LLMs to help manage the overwhelming volume of scientific literature.
CLEFT is a Contrastive Learning method designed for medical imaging that addresses the limitations of existing CLIP-like approaches that require extensive resources.
Sponsored
Gem's 2024 Talent Acquisition Trends report dives straight into the chaos. Why are 55% of companies doubling down on AI and automation [11]? Because they're tired of wasting time on mundane tasks. How are 68% of recruiters boosting their efficiency? By using smarter tools, not working harder. Want to know more? Read the full report here [11]to get ahead and stay ahead.
Open source text-to-video model that is close to the performance of closed video generation models.
Incredible vision language model that can almost run in real time. It outperforms closed models on a number of benchmarks.
RecDiffusion is a framework that enhances the process of making stitched images visually pleasing without cropping or distorting them.
After eight years on Google's robotics team, Alex Irpan has transitioned to the AI safety team at Google DeepMind, driven by a mix of personal reasons and a belief in the importance of addressing safety challenges as AI systems advance towards superhuman capabilities. They express concerns about the reliability of current AI safety measures, the increasing risks of unchecked AI development, and their commitment to contributing to AI safety, despite the complexities and controversies surrounding the field.
Despite U.S. restrictions aimed at preventing the export of AI chips to China, an active black market thrives, allowing sizable transactions like a reported $103 million deal involving Nvidia chips. Efforts to circumvent bans involve creating new companies, deceptive shipping practices, and relying on international distribution gaps. The underground trade persists, catalyzing debates over the effectiveness of U.S. export controls and their impact on American tech companies versus Chinese competitors.
This in-depth analysis of OpenAI's GPT-5 explores its potential improvements in reasoning, agents, and multimodality while considering OpenAI's technical and business challenges in the competitive AI landscape. It analyzes GPT-5's anticipated advancements and market expectations, forecasting that while GPT-5 will enhance existing capabilities, it may fall short of becoming a fully autonomous AI agent. OpenAI's holistic approach is expected to push the boundaries of large language models and maintain its influence in the AI industry.
Nvidia's upcoming Blackwell AI GPUs are delayed from Q4 2024 to potentially Q1 2025, impacting large customers like Microsoft and Meta.
OpenAI is considering releasing a tool that detects assignments written by ChatGPT but is cautious due to risks and potential unfair impact on non-English speakers.
Meta's CEO Mark Zuckerberg recently introduced Segment Anything 2 (SA2), an ML model for video segmentation that demonstrates notable progress from last year's image-only predecessor.
Sponsored
Research by Incogni [4] reveals that imposter calls (impersonating a person, agency, or company) were a major problem in 2023.
Groq, the lightning-fast AI chip startup, is raising a substantial round to meet demand for large language model inference.
Elon Musk has revived a lawsuit against OpenAI and its CEO Sam Altman, accusing them of fraudulently obtaining $44 million in seed funding by promising to keep OpenAI's technology open-source and prioritize the public good, only to later turn it into a for-profit venture with ties to Microsoft, which Musk claims betrays the original mission and has caused irreparable harm to both his interests and the public.
Greg Brockman is taking an extended vacation and John Schulman has gone to Anthropic to be an independent contributor.
Meta has announced a program to award groups using its models for good with $2m to help develop these tools for economically and socially impactful projects.
Researchers have developed a new method that uses language embeddings to improve point tracking in long video sequences.