sample efficient
How much a model (or human) learns from a given amount of data; high sample efficiency means learning a lot from few examples.
As defined in Sholto Douglas & Trenton Bricken — How LLMs actually think
Where it comes up · 3
- John Schulman (OpenAI Cofounder) — Reasoning, RLHF, & plan for 2027 AGI3 claims
- Sholto Douglas & Trenton Bricken — How LLMs actually think2 claims
- Why I don’t think AGI is right around the corner1 claim
Most important claims · 6 of 6Search all claims for “sample efficient” →
- Even if AI-run companies look better most of the time, they might have higher tail risk because AIs remain less sample-efficient in wacky situations and are more likely to malfunction in a big way; such practical considerations—plus liability and accountability rules—may push toward keeping humans in the loop for the near future regardless of average performance.John Schulman (OpenAI Cofounder) — Reasoning, RLHF, & plan for 2027 AGI
- In an eval, a model learned an actual human language (esoteric enough not to be in training data) in-context better than a human expert could over months, demonstrating superhuman in-context ingestion of information even though the base model knew nothing of the language beforehand.Sholto Douglas & Trenton Bricken — How LLMs actually think
- There is no large pretraining corpus of multimodal computer-use data, and just as the internet text corpus cracked NLP but wasn't enough for reliable agents (analogous to trying to train GPT-4 on only 1980's text data), it's unclear that text-only training gives a good prior over UIs or that RL fine-tuning is sample-efficient enough; there's no public evidence these models got less data hungry in unpracticed domains.Why I don’t think AGI is right around the corner
- Giving a model millions of tokens of context (e.g. a code base) produces a step-up in intelligence equivalent to large increases in model scale, because it solves the onboarding problem instantly without retraining—throwing in context dramatically improves next-token prediction the way scale normally would.Sholto Douglas & Trenton Bricken — How LLMs actually think
- No one has a good explanation for the scaling law with parameter count, but a sketchy account is that a bigger model is like a larger library of parallel, composable computations/circuits (analogous to mixture models or multiplicative weight updates with learned gating), so more width and depth give more chances that some circuit guesses correctly and gets upweighted, making bigger models more sample-efficient.John Schulman (OpenAI Cofounder) — Reasoning, RLHF, & plan for 2027 AGI
- As models get better they become more sample-efficient at recovering from errors and edge cases—rather than needing tons of data, a little data or generalization from other abilities lets them get back on track, whereas current models get stuck and lost.John Schulman (OpenAI Cofounder) — Reasoning, RLHF, & plan for 2027 AGI