Search the archive

GOOGLE CLOUD LAUNCHES NEW AI TOOLS FOR RETAILERS

AMAZON’S ALEXA GETS NEW GENERATIVE AI-POWERED EXPERIENCES

At CES, Amazon showcased new generative AI-powered Alexa experiences from developers Character.AI, Splash, and Volley. The experiences enhance the smart assistant's interactivity and creativeness.

MONARCH MIXER 32K RETRIEVAL MODEL

Together Compute has been exploring many different Transformer alternatives. It has released a retrieval model that outperforms many closed embedding models for retrieval tasks.

LANGUAGE MODELS ARE SENSITIVE TO PROMPTS

It turns out that prompting models for few shot benchmarks can give anywhere from 4-88% accuracy on the task, depending on the prompt. This research shows how to scientifically improve your prompts.

DEEP DIVE INTO BLIND MOTION DEBLURRING

This paper presents a thorough overview of deep learning's role in blind motion deblurring, a key area in computer vision. It covers everything from basic concepts and limitations of traditional methods to a detailed comparison of modern techniques like CNNs, GANs, RNNs, and Transformers.

A NEW METHOD FOR SINGING VOICE SYNTHESIS

Researchers have developed a new framework to better understand and analyze singing voices. By using self-supervised learning on isolated vocal tracks and focusing on out-of-domain generalization, they made progress in tasks like singing voice similarity and synthesis, improving upon existing technologies.

IMPROVING CODE TRANSLATION WITH SIMPLE LANGUAGE HELP

This study demonstrates that adding simple language explanations to code significantly improves how language models translate between 19 different programming languages, especially for complex codes.

MINIMA, A NEW EFFICIENT LANGUAGE MODEL

This project tackles the challenge of distilling large language models into smaller student models. The researchers discovered a consistent optimal capacity gap for different student models and used the insight to create MiniMA, a 3B model distilled from a 7B teacher model. MiniMa sets new benchmarks in efficiency and performance.

33B PARAMETER CYBERSECURITY LANGUAGE MODEL RELEASED

This model has been trained in both offensive and defensive security. It is a general purpose coding model that can assist in cyber security tasks. This means you can use it to secure your networks and learn how to protect from different hacks and vulnerabilities.

SPECTRAL PROPERTIES OF KERNEL MATRICES

This post explores how eigenvalues change for different Kernel Matrices and what that means for learning properties.

NVIDIA’S CEO ON LEADING THROUGH THE A.I. REVOLUTION (25 MINUTE PODCAST)

NVIDIA CEO and co-founder Jensen Huang discusses on this podcast how he leads his company in the face of accelerating change, sharing insights on how founders can integrate AI into their business in order to not be left behind.

NOUS RESEARCH RAISES $5M SEED

Nous Research, creators of the best open fine-tuned models available, have raised seed funding to continue building models.

IT’S HUMANS ALL THE WAY DOWN

People think that AI is going to replace a lot of jobs because everyone thinks that everyone else's job is easy. The desire to remove humans from the equation is based on ignorance. Even with the wildest of innovations, people still matter, and this can't be avoided. People want to appeal to other humans and be seen and understood.

KLART AI (PRODUCT)

AI assistant for everyone in your organization.

YOUR REFERRAL LINK - https://tldr.tech/ai?ref=2613930

OPENAI'S GPT STORE IS LIVE

OpenAI has launched the GPT store, which allows developers to get paid by building these agents. The company plans to feature GPTs every week.

REGULATORS AREN’T CONVINCED THAT MICROSOFT AND OPENAI OPERATE INDEPENDENTLY

European Union regulators are scrutinizing Microsoft's investment in OpenAI under EU Merger Regulations, concerned about potential antitrust risks and market distortion. Microsoft denies controlling OpenAI, despite its $13 billion investment and having an observer role on OpenAI's board. EU and UK authorities, along with the US FTC, are investigating the implications of such partnerships on competition and market dynamics.

UNSLOTH ACCELERATED MODELS IN TRL

Unsloth is a lightweight library for speeding up language model fine-tuning. It now works easily with TRL on common model architectures.

MASKED AUDIO GENERATION WITH A SINGLE NON-AUTOREGRESSIVE TRANSFORMER

Most audio generation techniques use diffusion or an auto-regressive model to generate sounds. This research doesn't use multi-steps or a complicated Transformer. Instead, it uses a masked language model over audio tokens.

KNOWLEDGE GRAPH GRAPH CONSTRUCTION WITH TECHGPT-2.0

TechGPT-2.0 enhances large language models for specific tasks like knowledge graph construction. The project, with its focus on named entity recognition and relationship triple extraction, also marks a significant stride in the Chinese open-source AI community.

DPO DATASET USING DISTILABEL

A new method that enables filtering of high-quality pair preferences for alignment. It improves baseline model performance by a substantial amount.

CHATBOT UI

An open-source repository that allows anyone to make a chat UI as part of their application.

MAPAI COMPETITION FOR BUILDING SEGMENTATION

MapAI launches an AI competition focused on building segmentation using aerial and laser data. This challenge, set for fall 2022 and concluding at the Northern Lights Deep Learning conference, invites participants to tackle two tasks: one using only aerial images and the other incorporating LiDAR data.

YOUR PRIVATE AI CAN HAVE EYES

Ollama now supports vision models. This means you can run Llava on your MacBook Pro for state-of-the-art vision and language performance.

AI AND THE FUTURE OF SAAS

AI is the juggernaut we all know and love these days. Its impact on SaaS has yet to be determined, so how do you go about mapping a path forward with a new or existing venture? Buckle up and give this a read (or listen).

BENCHMARKING GPT-4 TURBO

GPT-4 successfully completed 70% of programming tasks, while GPT-4 Turbo trailed slightly at 68.8%. Interestingly, GPT-4 Turbo required more second attempts, suggesting it might lack GPT-4's memorization strength. This was confirmed by a subsequent test.

smarter over time. Try it for free

OPENAI DEBUTS CHATGPT SUBSCRIPTION AIMED AT SMALL TEAMS

OpenAI is launching ChatGPT Team, a new subscription plan for its AI-powered chatbot that targets smaller teams of up to 149 members. This plan offers a dedicated workspace, access to OpenAI's latest models (including GPT-4 and DALL-E 3), tools for file analysis, and the ability to create custom AI applications without coding experience. ChatGPT Team is priced between $25 and $30 per user per month.

VALVE NOW ALLOWS THE “VAST MAJORITY” OF AI-POWERED GAMES ON STEAM

Valve has revised its policies to permit games that use AI tools on Steam. It will require developers to disclose the use of AI, whether for pre-generated content during development or live-generated content in-game.

POTIS AI (PRODUCT LAUNCH)

Potis uses AI to create automated behavioral interviews.

YOUR REFERRAL LINK - https://tldr.tech/ai?ref=2613930

RABBIT FOUNDATION MODEL AND COMPUTER (25 MINUTE VIDEO)

Rabbit created a large action model (LAM) that goes with its R1 pocket companion computer. The company's R1 device is a reimagination of the computer and smartphone, powered almost entirely by its LAM.

OPENAI’S NEWS PUBLISHERS DEALS REPORTEDLY TOP OUT AT $5M A YEAR

OpenAI is offering news publishers between $1M and $5M a year to use their content to train its AI models.

INTEL IS BRINGING THE AI PC TO THE CAR

Intel has announced an AI-focused chip for the automotive sector. Described as "the first-generation AI-enhanced software-defined vehicle system-on-chip," the announcement marks a strategic shift towards AI PCs and automotive AI applications.

A MINIMALIST APPROACH TO ALIGNMENT

Google researchers have introduced Self-Play Preference Optimization (SPO), a simpler form of alignment than traditional RLHF. The researchers turned to game theory and figured out single-player self play dynamics that are robust to noisy preferences and give strong performance.

MIXTRAL OF EXPERTS PAPER

The Mixtral paper has been released. There is no discussion of pretraining datasets. Most of the things disclosed are well known, but there is an interesting new insight on the discussion of routing evaluation between experts.

EVALUATION METRIC FOR TEXT-TO-3D MODELS

Researchers have developed a new evaluation metric for text-to-3D generative models, addressing the limitations of current single-criterion metrics. This advanced method uses GPT-4V to create prompts and compare 3D assets. It aligns closely with human preferences and offers versatility by adapting to various user-defined criteria.

AI TOOLKIT

AI Toolkit is a header-only C++ library that provides tools for building the brain of your game's NPCs.

HUMAN COMMUNICATION SIMULATION WITH MULTI-MODAL LLMS

SpeechAgents is a multi-modal AI system capable of simulating human communication with remarkable authenticity. This system, powered by a multi-modal LLM, can handle up to 25 agents. It shows potential in creating dramas and audio novels through its ability to mimic human dialogues, complete with consistent content, realistic rhythms, and expressive emotions.

VISION TRANSFORMERS WITH SELF-EMERGING TOKEN LABELING FOR ROBUSTNESS

In a significant advancement for Vision Transformers (ViTs), researchers have improved the robustness of the Fully Attentional Network (FAN) models using a new self-emerging token labeling (STL) framework. This approach involves training a FAN token labeler to create meaningful patch token labels, followed by training a FAN student model.

DOES AI RISK “OTHER” THE AIS?

This essay explores Robin Hanson's critique of AI risk discourse, focusing on the concept of "othering" AIs and the ethical implications of controlling or altering future AI and human values. Hanson argues that fearing AI as an "other" is a form of bias. Hanson's views may not sufficiently consider the risks of uncontrolled AI development and the challenges in aligning future AI values with human ethics.

ONE-YEAR ANNIVERSARY OF CHATGPT. HAS AI BECOME THE NEW TECH PLATFORM?

This article introduces the "Anatomy Framework", a tool for assessing the disruptive potential of any innovation, including AI. It explores innovation from five layers: underlying infrastructure, core platform, applications, tools, and ecosystem enablers. It also discusses the role of startup and incumbent innovators and the medium of innovation (hardware vs. software).

ENHANCING MEDICAL EXPERTISE IN LLMS

Researchers have developed a new Multi-disciplinary Collaboration framework to improve Large Language Models' understanding and reasoning in the medical field. This approach involves AI agents role-playing working together to analyze, discuss, and reach consensus on medical topics.

HUMAN POSE AND BODY SHAPE EDITING

This project introduces a one-shot method for editing human images that can make significant changes in pose and body shape while preserving the person's identity.

TRELLUS (PRODUCT)

A personalized AI sales coach for each and every cold call.

GOOGLE OPEN SOURCES 1.6T MOE MODEL (HUGGINGFACE HUB)

The switch transformer from Google was one of the first successful Mixture-of-Experts models. It now is available with code on the HuggingFace platform.

YOUR REFERRAL LINK - https://tldr.tech/ai?ref=2613930

OPENAI AND JOURNALISM

OpenAI addressed The New York Times lawsuit by emphasizing collaboration with news organizations, claiming that AI training with public content is fair use, committing to eliminating rare content regurgitation in their models, and expressing surprise over the lawsuit, given their ongoing efforts to address concerns.