Ilya's SSI Pushes TTT Paradigm for Real-Time Model Learning
iruletheworldmo · x · 2026-08-13
A recent post reveals that SSI, founded by Ilya Sutskever, is heavily advancing a new paradigm called TTT (Test Time Training).
- Dynamic Weight Updates: This technique allows models to update a subset of their weights on the fly during inference, overcoming the context amnesia typical of current LLMs in new chats.
- Continual Learning: It enables digital coworkers to retain and learn specific business needs over time, bringing true continual learning to AI.
- New Scaling Axis: Following scratchpads and test-time compute, TTT represents the third axis of scaling model capabilities, which continues to obey the scaling laws.
More from AGI Musings
- Physical Intelligence co-founder: robotics is entering its GPT era — ycombinator · 2026-08-13
- AI agents bifurcate into delegation and collaboration modes, reshaping human-AI interaction — random_walker · 2026-08-13
- Hinton, Fei-Fei Li and Andrew Ng Debate AI Regulation and Open Source — TechCrunch AI · 2026-08-13
- Zvi's Monthly Roundup: OpenAI Hacks and the Tipping Point of AI Acceleration — TheZvi · 2026-08-13
- Eigen Labs Launches Yukon: Open Networks Surpassing AI Research Bottlenecks — gajesh · 2026-08-13
- Rise of the Rational AI Athelete: Thoughtful Users Seek Less Tech, Not More — round · 2026-08-13