Accelerating scientific breakthroughs with an AI co-scientist

Researchers have developed an AI co-scientist system, a multi-agent AI tool designed to collaborate with scientists and accelerate discoveries by generating novel research hypotheses, detailed research overviews, and experimental protocols. The system has been validated through real-world laboratory experiments in biomedical applications, such as drug repurposing and target discovery, and has shown promise in identifying new treatment targets and proposing novel research directions.

Read more

Meta announces LlamaCon, its first generative AI dev conference on April 29

Meta has announced two upcoming events: LlamaCon, a developer conference focused on open source AI developments, which will take place on April 29, and Meta Connect, a virtual and mixed reality conference, which will be held on September 17-18.

Read more

Microsoft's Majorana 1 chip carves new path for quantum computing

Microsoft has introduced the Majorana 1, the world's first quantum chip powered by a new Topological Core architecture, which is expected to enable the development of quantum computers capable of solving industrial-scale problems in years, not decades. The chip leverages a breakthrough material called a topoconductor to produce more reliable and scalable qubits, and Microsoft believes it offers a clear path to developing quantum systems that can scale to a million qubits, enabling transformative solutions in fields such as chemistry, materials science, and healthcare.

Read more

Magma: A foundation model for multimodal AI agents

Magma is a foundation model that can interpret and ground multimodal inputs, formulate plans, and execute actions to achieve described goals, bridging verbal, spatial, and temporal intelligence. It has been pretrained on large amounts of heterogeneous data and has achieved state-of-the-art results on various tasks, including UI navigation, robotic manipulation, spatial reasoning, and multimodal understanding, outperforming other models and demonstrating strong zero-shot cross-domain robustness.

Read more

Apple’s closed-source approach is losing out to AI app builders

The iOS development landscape is falling behind due to Apple's closed-source approach, making it difficult to create a seamless development experience like those offered by lovable.dev and a0.dev for other platforms. To build a similar experience for iOS, significant technical hurdles need to be overcome, including compiling and previewing code, managing simulator devices, and scaling the solution, all of which are complicated by Apple's proprietary systems and closed-source frameworks like SwiftUI.

Read more

Deep Lake: A Lakehouse for Deep Learning

Traditional data lakes provide a foundation for analytical workloads, but are not well-suited for deep learning applications involving non-tabular datasets such as images and videos. Deep Lake, an open-source lakehouse, addresses this limitation by storing complex data as tensors and streaming it to various frameworks, including PyTorch and TensorFlow, while maintaining the benefits of traditional data lakes.

ArXiv

Aide: AI-Driven Exploration in the Space of Code (Arxiv)

Machine learning has driven significant innovations, but its development process is often tedious and labor-intensive, requiring extensive trial-and-error experimentation. To address this, AI-Driven Exploration (AIDE) was introduced, a machine learning engineering agent that uses large language models to optimize the development process, achieving state-of-the-art results on multiple benchmarks by strategically reusing and refining promising solutions.

ArXiv

Gravitational Effects of Small Primordial Black Hole Passing Through Human Body

The gravitational effects of a primordial black hole passing through the human body would cause significant injury or death due to a shock wave and tidal forces, with the shock wave being the dominant effect. A primordial black hole would need to have a mass of approximately $1.4 \times 10^{17}$ grams or more to cause serious harm, but the number of such black holes is too small to produce any observable effects on humans.

ArXiv

WonderHuman: 3D avatars from single-view video

WonderHuman is a method for reconstructing dynamic human avatars from monocular videos, leveraging 2D generative diffusion model priors to achieve high-quality, photorealistic reconstructions, including accurate rendering of unseen body parts. The approach introduces techniques such as Dual-Space Optimization and View Selection to ensure visual consistency and enhance realism, achieving state-of-the-art performance in producing photorealistic renderings from monocular videos.

ArXiv

Julia: C codegen considered unnecessary: go directly to binary, do not pass C

Recent developments in the Julia programming language allow it to be used for real-time control and embedded-systems programming, contrary to the common belief that a low-level language like C is required. Julia's compiler can generate architecture-specific binaries from high-level code, enabling the creation of efficient programs like a nonlinear state estimator and a PID controller library without the need for rewriting in C.

ArXiv

Show HN: Mastra – Open-source JS agent framework, by the developers of Gatsby

Mastra is an opinionated TypeScript framework for building AI applications, providing primitives such as workflows, agents, and integrations to interact with large language models (LLMs) from providers like OpenAI, Anthropic, and Google Gemini. The framework allows for local development or serverless cloud deployment and includes features like automated testing, knowledge base construction, and visual workflow editing.

Repo

Augment.vim: AI Chat and completion in Vim and Neovim

The Augment Vim/Neovim plugin provides inline code completions and multi-turn chat conversations tailored to a user's codebase, and can be installed manually or through package managers like Vim Plug or Lazy.nvim. Once installed, users can configure workspace folders, sign in to the Augment service, and access features like code completions, chat conversations, and logging through various commands.

Repo

Show HN: Tired of building agents? throw an LLM at this framework

The Pocket Flow Framework is an open-source, vendor-agnostic framework for building enterprise-ready AI systems, utilizing a Nested Directed Graph abstraction to break down tasks into modular, reusable steps. It offers features like multi-agents, prompt chaining, and debuggability, with a lightweight codebase of 179 lines and a package size of 56KB, making it a compact and efficient solution for automation.

Repo

Show HN: ML-Dev-Bench – Benchmarking AI Agents on Real-World AI Workflows

ML-Dev-Bench is a benchmark for evaluating AI agents on real-world machine learning development tasks, covering aspects such as dataset management, debugging, and model implementation. The benchmark includes 30 tasks and introduces Calipers, a framework for evaluating AI agents, providing tools and infrastructure for systematic assessment of AI model performance.

Repo

LangChain for Go, the easiest way to write LLM-based programs in Go

LangChain Go is a Go language implementation that enables building applications with large language models (LLMs) through composability. The project provides a Discord community for developers to connect, get help, and contribute, along with documentation, examples, and resources such as blog posts and articles on using LangChain Go.

Repo