Cape is now part of the MPC Alliance. We’re thrilled to join these industry pioneers on the mission to accelerate awareness and adoption of secure multiparty computation (MPC) technology. Together we will improve #dataprivacy and #security efforts. #secureMPC #encryptedlearning

We are hiring cryptographers at @capeprivacy! Come join us and be part of an awesome and fully remote team capeinc.bamboohr.com/jobs/view.php?…

Another big step in our journey! 🚀 Our new self-service platform for running #AI predictions in SnowflakeDB is live. #Financialservices organizations can now use #encrypted data for powerful predictive modeling safely in the #cloud. Read more here: globenewswire.com/news-release/2…

::Introducing the Cape API:: Keep sensitive data private while prompting LLMs like GPT-4 and GPT 3.5 Turbo. Easily de-identify sensitive data like financial, legal, and internal docs before sending to OpenAI Try the playground free: chat.capeprivacy.com How? /🧵

Making progress on privacy! “California lawmakers pass Delete Act that would force data brokers to eliminate all personal info they possess if people request it” fortune.com/2023/09/15/cal…

Llama3 was trained on 15 trillion tokens of public data. But where can you find such datasets and recipes?? Here comes the first release of 🍷Fineweb. A high quality large scale filtered web dataset out-performing all current datasets of its scale. We trained 200+ ablation

i'm having a super fun time collaborating with Eugene Yan, Bryan Bischof fka Dr. Donut, Charles 🎉 Frye, Hamel Husain, & jason liu on a 3-part series on working with LLMs. i learned so much from them that i really think it's the best resource on applied LLMs. here's part 1: oreilly.com/radar/what-we-…

This talk by Ben Clavié is the highest value per second talk I have ever watched on RAG Chapter summaries and additional links in next tweet

Among the most impressive aspect of the Llama 3.1 release is the accompanying research paper! Close to 100 pages of deep knowledge-sharing on LLMs like we havn't seen very often recently What a treat! It covers everything, pretrainining data, filtering, annealing, synthetic

It’s Sunday morning we have some time with the coffee so let me tell you about some of our recent surprising journey in synthetic data and small language models. This post is prompted by the coming release of an instant, in-browser model called SmolLM360 (link at the end) The

I'll get straight to the point. We trained 2 new models. Like BERT, but modern. ModernBERT. Not some hypey GenAI thing, but a proper workhorse model, for retrieval, classification, etc. Real practical stuff. It's much faster, more accurate, longer context, and more useful. 🧵

A few weeks ago, OpenAI announced Reinforcement Fine-Tuning (RFT)—a new way to adapt LLMs to complex tasks with very little training data. Here’s a quick rundown of how it works, why it’s a big deal, and when you should use it. 🧵

The RLHF method behind the best open models! Both DeepSeek and Qwen use GRPO in post-training! Group Relative Policy Optimization. GRPO was introduced in the DeepSeekMath Paper last year to improve mathematical reasoning capabilities with less memory consumption,

After 6+ months in the making and burning over a year of GPU compute time, we're super excited to finally release the "Ultra-Scale Playbook" Check it out here: hf.co/spaces/nanotro… A free, open-source, book to learn everything about 5D parallelism, ZeRO, fast CUDA kernels,