Sublime
An inspiration engine for ideas
Welcome to the Era of Experience
The text discusses a new AI era focused on superhuman agents learning primarily from their autonomous, continuous interactions with the real world, surpassing prior human-data-based models by leveraging experiential learning and grounded rewards.
storage.googleapis.com
Rich Sutton just published his most important essay on AI since The Bitter Lesson: "Welcome to the Era of Experience"
Sutton and his advisee Silver argue that the “era of human data,” dominated by supervised pre‑training and RL‑from‑human‑feedback, has hit diminishing returns; the future will belong to agents... See more
fresh-eggs-flying-lessons
Tips and insights for brands and marketing: Avoid common mistakes; think differently; understand customer needs deeply; don't rely solely on data; strive for loyalty and customer satisfaction.
Link
the ‘multi-armed bandit’, and it's the textbook example of the Explore/Exploit trade-off. The optimal solution is to start out by exploring as much as possible, and gradually move towards exploiting as your time runs out.
Richard Meadows • Optionality: How to Survive and Thrive in a Volatile World
ה-Bitter Lesson של AI Research.
ריצ׳ארד סוּטוֹן הוא אחד מאבות ה-Reinforcement Learning וחתן פרס טיורינג לשנת 2025 על תרומתו לתחום. ב-2019 הוא כתב טור סופר חשוב בשם The Bitter Lesson. זה קרה לפני ChatGPT ולפני StableDiffusion. מהו השיעור המר?
>>
Nir Ben-Zvix.com
Agent Design Pattern: Parallel Rollouts
Inspired by Tree-of-Thought [1] and @corbtt's Universal Reward Function [2], lately I've been using a best-of-n pattern (dubbed “parallel rollouts” internally) and seeing consistently strong results
When designing an agent, retry-on-failure is... See more
Okay okay, spent my weekend gooning around learning GRPO math. Here's some takes.
Essentially, this is me yapping through a recap of smaller details on how GRPO is implemented, what Dr. GRPO changes, why, DAPO, connections to PPO, aggregating batches...
Reading list below.... See more
Nathan Lambertx.com
This is fascinating: Rich Sutton on the "bitter lesson" of AI research: https://t.co/LW5TOGTIKw https://t.co/z4nsRbH01V