🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Apple joins the #LLM race: their OpenELM model uses a scaled-attention mechanism for more efficient parameter allocation and outperforms similarly-sized models while requiring fewer tokens for training. Read more in my latest InfoQ news!

OpenAI releases their newest model: GPT-4o...it's faster and has improved capabilities in handling speech, vision, and multilingual tasks. Read more in my latest InfoQ news!

The Stanford AI Index report is out, so you can keep up with top trends in AI, such as 8x growth in Generative AI investment since 2022. Read more in my latest InfoQ news!

OpenAI has published their Model Spec that describes rules and objectives for the behavior of their #GPT models. It's intended for use in creating data for fine-tuning the models. Read more in my latest InfoQ news!

One caveat with most #LLMs is the context-length limit: you can only input so much data, or only have a limited-length conversation with a chatbot. Meta's new LLM, MEGALODON, addresses this with an unlimited context length. Read more in my latest InfoQ news!

No InfoQ news from me this week, BUT! We have published our Generative AI e-magazine! I'm very excited about this one, and I hope you enjoy it. Check it out!

#ChatGPT is pretty good at writing code, but it's not perfect. OpenAI created CriticGPT to help find bugs in ChatGPT-generated code. CriticGPT catches more bugs and produces better critiques than human coders. Read more in my latest InfoQ news!

Google's new open-source #LLM, Gemma 2, outperforms other models of comparable size and is competitive with models 2x larger. Read more in my latest InfoQ news!

OpenAI's latest #ChatGPT update is out: GPT-4o mini. This is a smaller, faster, cheaper model, that outperforms GPT-3.5 turbo. Read more in my latest InfoQ news!

No joke: Google's JEST automates dataset curation, so that models trained on the data require 10x less computation than baseline methods. Read more in my latest InfoQ news!

Three new open-weight #LLMs from Mistral AI: Mistral NeMo, a 12B parameter general-purpose LLM; Codestral Mamba, a 7B parameter code-generation model; and Mathstral, a 7B parameter model fine-tuned for math and reasoning. Read more in my latest InfoQ news!

University of Pennsylvania Researchers Develop Processorless Learning Circuitry bit.ly/3SOHWtX by Anthony Alford

InfoQ AI, ML, and Data Engineering Trends in 2024 bit.ly/3X207yI by srinip, Roland Meertens, Anthony Alford, Daniel Dominguez, Mandy Gu, Namee Oberst

Kolmogorov–Arnold Networks (KAN) models, a new type of neural network, outperform larger perceptron-based models on physics modeling tasks and provide a more interpretable visualization. Read more in my latest InfoQ news!

Maybe you've heard about #AppleIntelligence (AI? well played, Apple!). Read my latest InfoQ news about the Apple Foundation Models that power several of its features.

Alibaba released Qwen2-Math, a series of LLMs tuned for solving mathematical problems; and Qwen2-Audio, a family of multi-modal LLMs that can accept voice or text input. Read more in my latest InfoQ news!

#Doom enthusiasts know the game will run on just about anything. Now Google has it running in a neural network. Read more in my latest InfoQ news!

Can an AI coding assistant help you? A recent study suggests developers could increase productivity by 26%. Read more in my latest InfoQ news!

Introducing ChatGPT Search! ChatGPT can now incorporate current information from the web and include links to its sources. Read more in my latest InfoQ news.

Claude 4 is here! It outperforms other LLMs on coding benchmarks like SWE-bench. Read more at InfoQ