maadaa AI Data News: Stable Diffusion 3 Outshines in Text-to-Video Generation, TripoSR Transforms 2D to 3D, and More

June 24, 2024Updated 7:55 am

News Date: March 5th ~ March 11th, 2024

1. Elon Musk Says xAI Will Open-Source Grok This Week

Elon Musk has announced that xAI, his artificial intelligence company, plans to release Grok, an advanced AI model, as open-source software later this week. The decision aims to make AI technology accessible to developers worldwide, thus enabling them to contribute to and benefit from Grok’s capabilities. This move is seen as a significant step towards fostering innovation and collaboration in the AI field. Musk emphasized the importance of transparency and the collective advancement of AI technologies for the benefit of society.

2. Revolutionizing Text-to-Video Generation: Stable Diffusion 3 outperforms State-of-the-Art AI System

Stability AI has published a research paper on Stable Diffusion 3, a “text-to-video” AI model, which provides an in-depth description of the technical details. The paper is now available on arXiv.

Based on human preference evaluations, Stable Diffusion 3 outperforms state-of-the-art text-to-image generation systems such as DALL·E 3, Midjourney v6, and Ideogram v1 in typography and prompt adherence.

The Multimodal Diffusion Transformer (MMDiT) architecture uses separate sets of weights for image and linguistic representations, improving text comprehension and spelling compared to previous versions of Stable Diffusion 3.

Paper: https://stabilityai-public-packages.s3.us-west-2.amazonaws.com/Stable+Diffusion+3+Paper.pdf

3. Open-source 3D generative model TripoSR turns a single image into a 3D model in 1 second

Stability AI recently announced that it has partnered with VAST, a 3D generative modeling startup, to open source TripoSR, a fast 3D object reconstruction model that generates high-quality 3D models from a single image in less than 1 second. TripoSR runs on a low inference budget and is fully usable by users without GPUs.

When tested on an NVIDIA A100, it generates draft-quality 3D output (textured mesh) in about 0.5 seconds, outperforming other open image-to-3D models such as OpenLRM.The TripoSR model weights and source code are available for download under the MIT License, which permits commercial, personal, and research use.

Project: https://github.com/VAST-AI-Research/TripoSR

Report: https://stability.ai/s/TripoSR_report.pdf

4. AGI startup Ema has raised $25 million in funding

According to TechCrunch, Ema, a US-based generative AI startup, has announced that it has raised $25 million in funding and has already amassed several customers.Surojit Chatterjee, CEO and co-founder of Ema, said that their goal is to build a general-purpose AI workforce. The company has launched two products, the Generative Workflow Engine (GWE) and EmaFusion, a simulated human response tool that companies can use in applications ranging from customer service to internal employee productivity.

5. Microsoft’s AI-Powered PCs Are Here

Latest rumors suggest that Microsoft is about to launch its latest products: the Surface Pro 10 and Surface Laptop 6. These are so-called “AI PCs” that come equipped with fast Intel Core Ultra/Qualcomm Snapdragon X Elite processors and advanced AI capabilities.

These new devices allow you to search for information across documents, web pages, pictures, and chats using natural language. They also keep a searchable history of your activities, which can be used to provide you with intelligent suggestions. Additionally, you can perform text-based image editing with ease.

There are also reports that Microsoft is working on a new feature called “AI Explorer” for the upcoming Windows 11.

6. Anthropic’s Latest Language Models- Claude 3 Series: Fast, Affordable, and High-Performing

Anthropic has recently launched three state-of-the-art Language Models (LLMs) which are giving tough competition to other popular models like ChatGPT and Google Gemini.

The Claude 3 Haiku model is the fastest and most affordable model in the Claude 3 series. It is perfect for applications that require quick responses.

The Claude 3 Sonnet model is designed to deliver robust performance at a lower cost than its competitors, making it an ideal choice for enterprise-level applications. It is built to handle large-scale AI deployments.

The Claude 3 Opus model is the best of the three and can handle complex tasks with ease. It has been trained to perform well in open-ended scenarios and high cognitive processing tasks, making it almost as good as human understanding.

Related Open and Commercial Datasets:

maadaa.ai has also found some datasets related to this week’s news. Hope it helps. Stay tuned!

1. AI Photo-Video Editing Open Dataset

This open collection of datasets includes specialized fine segmentation datasets for precise object recognition and manipulation, human body segmentation for advanced body-based manipulation, face segmentation for realistic and personalized face manipulation, and more.

2. Fashion & e-Commerce Open Dataset

The dataset has 24 real-world scenarios with 33 precisely labeled sub-datasets. It's diverse and realistic and can be used for object detection, product recognition, personalized recommendations, virtual fittings, beauty AI, and more.

Citation:

Any further information, please contact us.