Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
This paper introduces meta-in-context learning, a recursive self-improvement process for large language models. By studying regression and decision-making tasks, the researchers demonstrate that meta-in-context learning enhances the models' in-context learning abilities, modifies their strategies, and achieves competitive performance on real-world problems. This work sheds light on improving language models without relying solely on traditional fine-tuning approaches.
ControlVideo is a novel framework that addresses the limitations of generating videos from text by introducing a training-free approach. By leveraging structural consistency, enhancing appearance coherence, mitigating flicker effects, and employing hierarchical sampling, ControlVideo outperforms existing methods in generating high-quality videos that are both short and long. Importantly, ControlVideo achieves these results efficiently, generating videos within minutes using a single NVIDIA 2080Ti GPU.
An open-source visual programming environment for battle-testing prompts to LLMs.
LoopGPT is a re-implementation of the popular Auto-GPT project as a proper python package, written with modularity and extensibility in mind.
Amazon’s cloud customers are clamoring to get their hands on the ChatGPT-style technology the company unveiled six weeks ago. But instead of being allowed to test it, many are being told to sit tight, prompting concerns the artificial intelligence tool isn’t fully baked.
The article discusses the challenges of using artificial intelligence to moderate content across multiple languages. It argues that current AI systems are not able to accurately detect harmful content in all languages, and that this can have serious consequences. It finishes by calling for more research into how to improve the accuracy of AI content moderation systems, and for more transparency from social media companies about how they are using AI to moderate content.
The White House said on Tuesday that it would ask workers how their employers use artificial intelligence to monitor them, as it allocates federal investments in the technology, which is expected to change the nature of work.
Transformers displaced recurrent neural networks as the primary tool for most text based tasks. However, these Transformers are expensive and limited in context due to the architecture. RWKV is an attention variant (vaguely reminiscent of linear attention) that claims to have the best of both Transformers and recurrent models. The Discord is quite active, but there hasn't been a formal write up of this hyped model. This paper is the first to explain the model in depth.
Leveraging the power of language processing models, the newly proposed VideoLLM system converts all video inputs into a type of language that can be analyzed more efficiently. By successfully testing on multiple datasets, VideoLLM proves that it can handle various tasks, suggesting that language model's reasoning abilities can be effectively used for understanding and analyzing video content.
DB-GPT is an experimental open-source project that uses localized GPT large models to interact with your data and environment.
Seamlessly integrate powerful language models like ChatGPT into scikit-learn for enhanced text analysis tasks.
An op-ed by Google CEO Sundar Pichai in which he discusses how Google is building AI responsibly, with a specific focus on developing guardrails, creating smart regulation, and fostering international cooperation.
Microsoft has agreed to sign all AI art that its apps generate with a cryptographic watermark indicating it was made with an algorithm.
Buzzfeed continues their lean into AI with Botatouille, an AI kitchen assistant.
Girlfriend GPT is a Python project to build your own AI girlfriend using ChatGPT4.0.
Sponsored
Researchers built a powerful tool called mLongT5 that can understand and translate long texts in multiple languages. It outperforms other similar tools in tasks like summarizing text and answering questions in various languages.
Researchers have developed a tool named DragGAN, which allows users to modify images precisely by dragging any point in the image to a target location, enabling changes in pose, shape, expression, and layout. It outperforms existing image manipulation methods, proving its efficiency not only in creating realistic results but also in tracking changes, even when manipulating real images.
Authors have designed a new way to teach language models to understand and plan in the real world by using a virtual simulator to give them 'embodied experiences.' This new method significantly improves the models' abilities and makes smaller models perform as well as, or even better than, larger models like ChatGPT.
SuperAgent is a powerful tool that simplifies the configuration and deployment of LLM (Large Language Model) Agents to production.
Sponsored
Apple has restricted use of ChatGPT and other external artificial intelligence tools as it develops its own similar technology.
AI is revolutionizing healthcare with applications in predicting patient health outcomes, personalizing treatment, and automating tasks. However, these benefits come with challenges like data privacy, data requirement, and algorithmic bias that must be addressed for successful implementation.
Zapier is focusing on developing AI-powered automation tools to allow everyone to create and customize software applications. By integrating with AI tools such as OpenAI, Leap, and Vowel, Zapier aims to streamline operations and provide tailored software solutions. The company has launched features like the Zapier ChatGPT plugin and the Natural Language Actions API. Additionally, it announced two new AI beta features: creating a Zap using plain English and allowing users to customize workflows without writing any code.
DarkBERT is a new AI that's been trained specifically on the unique language used in the Dark Web, which is different from what you'd find on the regular internet. By comparing it to other AIs and looking at various use cases, we've found that DarkBERT is better at understanding and working with Dark Web text, making it a useful tool for future research in this area.
Symbol tuning is a new way to train AIs, where we replace normal language labels (like "positive" or "negative") with random symbols (like "foo" or "bar"), forcing the AI to learn from the connections between inputs and labels. This method improves the AI's performance on new tasks, helps it understand instructions better, and makes it more skilled at logical reasoning tasks, with significant improvements seen across different benchmarks.
Boundless DAS is a new method we've developed to better understand how large AI models, like Alpaca, work on the inside, by looking for understandable patterns in their behavior. We've found that Alpaca solves a simple number problem by using two understandable variables in a consistent way, no matter what the inputs or instructions are, which is an important step towards understanding how these complex AI models really work.
GGML is a framework for running 4-bit quantized models on the CPU. This means you can run really large models locally on your laptop. Starcoder is one of the very best open source program synthesis models. Scoring very well on the challenging OpenAI human eval benchmark. This pull request adds that model to the GGML framework which means you can run starcoder on commodity hardware.
An AI-powered “junior developer” programming assistant.
A set of numbers for LLM developers to know that are useful for back-of-the envelope calculations.
Sponsored
It’s almost impossible to keep track of everything going on with AI right now — but this podcast [https://link.chtbl.com/tldrnopriors] is a start. Join co-hosts Elad Gil and Sarah Guo as they ask the smartest people in the industry the most pressing questions, like: How far away is AGI? What markets are at risk for disruption? How will commerce, culture, and society change? What’s happening in state-of-the-art research? And more. Listen to the latest episode of No Priors.
Caryn Marjorie, a social media influencer, launched CarynAI, an AI chatbot leveraging GPT-4 API technology developed by OpenAI that replicates her voice, mannerisms and personality. For $1 a minute, fans can chat with CarynAI in an “immersive AI experience” that feels almost like speaking to Marjorie herself.