AI News & Launches

LLMD: A Large Language Model for Interpreting Longitudinal Medical Records is a paper that introduces a language model trained on a vast corpus of medical records to analyze patients' medical histories effectively. The model, named LLMD, is fine-tuned for tasks such as normalizing metadata and abstracting high-level representations, with an emphasis on patient care data over a decade-long period from up to 140 care sites. The study demonstrates that LLMD outperforms both general and domain-specific models, providing notable improvements in interpreting real-world patient data.

Unsloth AI has addressed a critical issue in Gradient Accumulation that has been affecting training and finetuning of language models (LLMs). The bug led to higher training losses for larger gradient accumulation steps. The Unsloth AI team formulated and implemented a fix, resulting in improved accuracy for gradient accumulation that aligns with full batch training. Updating Unsloth and using their fixed trainer, as advised, can significantly lower associated errors.

The GitHub repository 'Tamagotchi' by Barqawiz provides a collection of Tamagotchi characters designed to give AI assistants a unique identity, inspired by the classic virtual pet device of the same name. The project features characters that can be controlled manually or through AI assistance.

Kagi has introduced an AI Image Filter to improve image search results by downranking and labeling AI-generated images, providing users with the option to exclude them altogether. Users can also block specific websites featuring AI or low-quality imagery.

MotherDuck has announced a new SQL function called prompt(), which integrates small language models such as OpenAI's gpt-4o-mini into SQL data warehouses. The function is designed for text generation, summarization, and structured data extraction without separate infrastructure. It is now available in Preview for MotherDuck users to apply advanced natural language processing techniques more cost-effectively and with simplicity.

The AI Mathematical Olympiad Progress Prize is a Kaggle competition aimed at advancing artificial intelligence in solving complex mathematical problems. Participants develop models that can crack math olympiad questions, with a chance to win prizes and contribute to educational AI research.

Google has introduced new customization options for NotebookLM's Audio Overviews, allowing users to guide AI host emphasis and expertise level. The update paves way for NotebookLM Business, integrated with Google Workspace, with features geared towards business, university, and organization use, emphasizing data privacy and security.

Adobe's Project Turntable, revealed at the MAX conference, showcases an advanced AI-driven image rotation tool that allows full 3D rotation of 2D vector art while maintaining original design aesthetics. This innovative concept impresses with its ability to intelligently fill in unseen image details during rotation.

Smart Invoices is an AI-powered mobile application designed to simplify invoicing and financial management for business owners. It features advanced AI-driven technology for creating professional invoices quickly and offers tools like automatic tax & discount calculation, cloud storage, and customizable branding.

Token Counter is an online tool designed to count tokens for various AI models to manage API costs, estimate processing times, and adhere to model limits. It utilizes the recommended tiktoken library for OpenAI models for high accuracy and supports multiple languages.

CodeAnt's AI Code Review is an automated service that helps improve the quality of code by providing insights and suggestions for better coding practices. It leverages AI to analyze code and highlight areas that could be optimized or refactored for better performance and maintainability.

ImageKit AI offers an advanced image processing and digital asset management platform, utilizing AI and Generative AI for efficiency. Features include generative fill, smart cropping, background removal, shadow addition, on-the-fly image creation using text prompts, image variation generation, visual quality improvement, and AI-powered tagging and search within their DAM system.

Reiden.ai is a powerful productivity tool that acts like a personal keyboard shortcut copilot, aiming to save users time by suggesting efficient shortcuts across multiple applications. It features a real-time workflow analysis, customized recommendations, and promotes ergonomic efficiency, all while emphasizing user privacy.

Prismy is a GitHub-native, AI-powered localization platform designed to save hours and ensure high-quality translations for businesses seeking to ship their products in multiple languages more efficiently. Key features include tailored AI translations, GitHub conflict avoidance, and effortless integration.

PlayNode.app is an online platform designed to enhance creativity with the help of AI. It allows users to create workflows, manage tasks, and integrate with React Flow. Features include adding boards, saving, sharing, and clearing workflows. The best experience is on desktop or laptop.

Convo is an AI-moderated platform designed for qualitative user research. It offers the depth of one-on-one interviews combined with the breadth of survey scalability. Convo supports multilingual responses and provides real-time insights with its automatic data analysis.

Prisma Optimize provides an AI-driven analysis of database queries, offering actionable recommendations to improve performance. It supports insightful query metrics and integrations with popular databases, assisting in query optimization for faster application development.

Highperformr offers a comprehensive suite of AI-powered social media management tools ideal for startups and small to large businesses. It simplifies publishing, team collaboration, analytics, and audience engagement.

LOTUS is a state-of-the-art visual foundation model presented in arXiv 2024, using diffusion techniques for zero-shot dense geometry prediction tasks. Its novel approach and fine-tuning protocol allow it to excel in depth and normal estimation benchmarks with minimal training data, outperforming other methods in quality and efficiency.

A discussion on Hacker News explores the best practices for adding guard rails to LLM responses without disrupting streaming functionality. Participants share insights on using regex filters, LLAMA Guard, token sampling, and fine-tuning stop words for optimal performance.

Ichigo is an open-source, real-time voice AI project rebranded from llama3-s. It extends the text-based LLM's 'listening' capabilities, using an early fusion technique to combine text and audio inputs. The maintainers of Ichigo showcase ongoing progress, invite collaboration, and provide both tutorials and tools for engaging with the project.

Foyle is an AI system designed to make the operations of software deployment and management easier for developers without extensive knowledge in DevOps, by translating intent into executable commands within VScode notebooks.

GitHub repository 'qc_npu_benchmark' provides a code sample to benchmark models on Qualcomm's Windows PCs, specifically on Microsoft Surface tablets equipped with Qualcomm's NPU. The benchmarking reveals suboptimal NPU performance, achieving only a fraction of the claimed capabilities.

In a recent article, Pete Warden expresses his disappointment with the AI performance of Qualcomm's NPUs on Windows tablets. Despite the chips' promise of up to 45 trillion operations per second, Warden's benchmark tests using foundational AI operations only achieved 573 billion ops/sec, which is less than 1.3% of the advertised capability, and four times less than an Nvidia RTX 4080 in his gaming laptop. He speculates that the issues may stem from the software stack and hopes for future improvements through software updates.

Mistral AI announces world-class edge models, Ministral 3B and Ministral 8B, offering unparalleled efficiencies for on-device and at-the-edge applications. Setting benchmarks in knowledge, reasoning, and efficiency under 10B models, they are ideal for a range of use cases from smart assistants to autonomous robotics.

Wandpen is an AI writing assistant designed as a Chrome extension that provides distraction-free assistance for writing tasks. It offers users the ability to rewrite their emails and text with AI support, aiming for clarity and confidence in communication.

The website provides in-depth analysis and coverage of the political debate between Kamala Harris and Donald Trump, featuring articles, opinion pieces, and user discussions.

Tattoon is a website dedicated to everything related to tattoos, offering designs, care advice, inspiration, and artist showcases to enthusiasts and professionals alike.

FAQWidget.ai is an AI-powered service that generates a targeted FAQ widget for websites, aiming to drive engagement and increase sales. It offers instant generation, SEO-friendly content, no coding requirements, and options for customization and targeted keywords.

Prediko provides AI-assisted inventory management solutions for Shopify brands, offering tools like demand planning, insights, analytics, supply planning, raw materials management, and purchase order management. The services aim to streamline inventory KPI tracking and purchase decisions, minimizing stockouts and maximizing revenue.

Dash by Dropbox is an AI-powered platform that combines universal search, content organization, and access control. It connects with apps like Google Drive, Outlook, and Dropbox, to centralize work operations.

Packmind is a revolutionary platform designed to accelerate team learning and enhance engineering performance through AI integration and the promotion of best coding practices.

The National Archives is set to introduce 'Archie AI', a public-facing AI chatbot. Despite a ban on ChatGPT for posing a security risk, Google's Gemini AI has been promoted to enhance employee productivity. Employee concerns over AI's role in archiving accuracy, data security, and environmental impact arose during a presentation.

Meta showcases new open AI hardware designs including a new AI platform and advanced network fabrics at OCP Global Summit 2024. The initiative invites collaboration to drive innovation in AI, backed by Meta's development of infrastructure like Llama 3.1 405B model and 24K-GPU clusters.

OpenAI is reportedly transitioning from a nonprofit to a for-profit company, amidst a massive $6.6 billion funding round and a $157 billion valuation. Key executives, including CTO Mira Murati, have resigned. Doubts rise over OpenAI's sustainability as it faces enormous annual losses and a potential subprime AI crisis, despite ambitious revenue projections.

Command AI, formerly known as CommandBar, is being acquired by Amplitude. The product remains operational and will eventually integrate with Amplitude's stack to improve user experiences. The acquisition follows a partnership on a popular integration and aims to make software easier to use. The Commander AI team thanks investors, customers, and users for their support over 4.5 years.

Researchers have revealed methods for inserting invisible text into prompts and outputs of AI chatbots like Claude and Copilot, exploiting Unicode quirks to create steganographic channels. This allows for covert communications and potential data leaks.

A Hacker News user kchnkrml seeks recommendations for fine-tuning a Language Model (LLM) for documentation interaction to make information in their framework's documentation more accessible to new users. They would like the model to map plain text questions to documentation resources.

Talkstack.ai offers AI-powered customer support and sales agents, capable of complex tasks in multiple languages, without pre-recordings or trigger words. AI agents can provide reminders, support, marketing, and scheduling services, boasting 98% automation and enterprise-grade security.

Perplexity: Ask Anything is an AI-powered search and discovery app designed for Mac users. It offers features such as Pro Search, voice or text prompts, Thread Follow-Up, cited sources for answers, and a personal archive. It is available for pre-order and will be released on October 24, 2024.