Andrej Karpathy unveils nanochat, a full-stack training and inference implementation of an LLM in a single, dependency-minimal codebase, deployable in 4 hours
It provides a full ChatGPT-style LLM, including training, inference and a web Ui … X: Clem / @clementdelangue : Am I wrong in sensing a paradigm shift in AI? Feels like we're moving from a world obses...
A US judge rules Anthropic's use of copyrighted books to train AI was fair use, but its storage of pirated books in a central library for training LLMs was not
but it's still in trouble for stealing books Blake Brittain / Reuters : Anthropic wins key US ruling on AI training in authors' copyright lawsuit Jason Koebler / 404 Media : Judge Rules Training AI on...
The Wikimedia Foundation partners with Kaggle to release a dataset of “structured Wikipedia content in English and French” optimized for AI model training
Data science platform Kaggle is hosting a Wikipedia dataset that's specifically optimized for machine learning applications.
Apple plans to begin on-device, privacy-centric analysis of user data, comparing it to synthetic data, to improve AI, in the iOS 18.5 and macOS 15.5 betas
without reading them Emma Roth / The Verge : Apple's complicated plan to improve its AI while protecting privacy Ivan Mehta / TechCrunch : Apple details how it plans to improve its AI models by privat...
Anthropic's Claude 3.7 Sonnet reportedly cost “a few tens of millions of dollars” to train, similar to Claude 3.5 and cheaper than GPT-4, which cost over $100M
“Assuming Claude 3.7 Sonnet indeed cost just ‘a few tens of millions of dollars’ to train, not factoring in related expenses, it's a sign of how relatively cheap it's becoming to release state-of-the-...
Industry insiders say DeepSeek's focus on research makes it a dangerous competitor as it's willing to share breakthroughs rather than protect them for profits
China is pulling the same trick. — www.ft.com/content/747a... Mastodon: Brian Kung / @briankung@hachyderm.io : “There's a pretty delicious, or maybe disconcerting irony to this, given OpenAI's found...
YouTube adds a tool to let creators authorize third parties, including Amazon, Anthropic, Apple, Meta, Microsoft, and OpenAI to train AI models on their videos
Here's How It Works Chandraveer Mathur / Android Police : YouTubers can now choose if their videos are used to train AI Bluesky: @eshumarneedi.com : As if those third parties weren't already training ...
Sources: Apple and Baidu are working to add AI features to Chinese iPhones but have clashed over model training data, and the models have struggled with prompts
Apple and Chinese search giant Baidu are working together to add artificial intelligence features to iPhones sold in China …
Anthropic raised another $4B from Amazon, will make AWS its primary training partner, and is working with Annapurna Labs to develop new Trainium accelerators
from chips to software. Mady Mills / @madisonmills22 : $AMZN to invest another $4B into Anthropic. Big news in the release: Anthropic is now naming AWS its primary training partner...and will use AWS ...
Sources: Huawei's Ascend chips still lag far behind Nvidia's for model training and have stability issues, slower inter-chip connectivity, and inferior software
Tech group's Ascend artificial intelligence chips are being widely adopted but Chinese companies complain of performance problems