VisiveAI

DeepSeek open sources Super-Agent as a research prototype, a proactive coding agent it says can monitor group chats for coding issues and suggest solutions (@debbywuintaipei / Bloomberg) (Visit Techmeme dot com for the link and full context!)

Highly recommended paper from CMU and others on how foundation models get trained. It fundamentally asks whether pretraining, midtraining, SFT, and RL really need to be separate stages. They show that this separation isn't fundamental. All of these stages can run under a single likelihood-based algorithm, which they call unified training. The key step treats suitable RL objectives as maximum-likelihood fitting to model-generated targets. RL generations and static data then share one loss and one trainer. This changes the design problem. You now choose which data to train on, in what proportions and when. Staged training becomes one special case. They first trained a 124M model from scratch on reasoning puzzles. The best unified runs beat staged training on the RL tasks. They also beat it on larger puzzles the model never saw. Those runs reused the midtraining corpus up to 4 times. Staged training lost accuracy when midtraining ran past its best budget. Under a fixed compute budget, unified training had the best accuracy at the smallest and largest budgets. Staged training led at intermediate budgets. The real-world test used a 0.76B model on math RL, with 0% to 5% midtraining data mixed in. Math accuracy stayed similar while retention improved. Coding accuracy dropped as the mix grew. Continual learning gave the clearest result. After math training, all unified models beat staged training on final coding accuracy, despite starting lower. The 5% mix did best. The experiments are small-scale. The authors have not found a mixture or schedule that works best across the board. arxiv.org/abs/2610.40564 Chat with Paper: academy.dair.ai/papers/train…

The FBI is releasing information advising the Democratic People's Republic of Korea (North Korea) was responsible for the theft of more than $6 billion USD in virtual assets from cryptocurrency exchange, Binance, on or about October 29, 2026. The FBI refers to this specific North Korean malicious cyber activity as "TraderTraitor." The FBI, together with partners at the USSS and IRS-CI, has observed TraderTraitor actors converting stolen assets primarily into Ethereum (ETH). It is expected these assets will then be further laundered by converting them using DeFi and instant swap services as they prepare them for further laundering stages. How You Can Help: FBI, USSS, and IRS-CI are urging private sector entities including THORNode operators, RPC node operators, instant exchanges, DeFi services, blockchain analytics firms, and other virtual asset service providers to block transactions with or derived from addresses TraderTraitor actors are using to launder these stolen assets. A list of addresses associated with this activity is available here: fbi.gov/investigate/cyber/al…

Tests: chatbots rely on three obscure AI-built voter sites for US midterm answers; Copilot cited the error-prone Ballot Brief in 75% of tests and ChatGPT in 22% (Washington Post) (Visit Techmeme dot com for the link and full context!)