AGI
AGI is machine intelligence that matches human competence across the full range of tasks rather than excelling at one.
used by three
Concepts
Terms this publication defines and builds on. Each entry is versioned; posts cite the version they depended on.
Agent harness
An agent harness is the program wrapped around a language model that turns isolated completions into sustained work: it runs the loop, exposes tools, keeps and prunes context, and decides when to stop.
used by three
GRPO
GRPO — Group Relative Policy Optimization — is a reinforcement-learning algorithm that scores each attempt against the other attempts in its own group rather than against a learned value function: the model generates several independent tries at the same problem, and an attempt's advantage is how it did relative to its siblings.
used by one
On-policy distillation
On-policy distillation is teacher–student training in which the student learns from the teacher's judgment of the student's own attempts, rather than from the teacher's demonstrations.
used by one
RLVR
RLVR — reinforcement learning with verifiable rewards — is post-training in which the reward comes from a check the environment can run, not from a learned preference model or a human rating: a test suite passes, an answer matches, a program compiles.
used by two
Scaling laws
The empirical regularity that model capability improves predictably with scale — more compute, more data, more parameters, more intelligence out.
used by two
Supervised fine-tuning
Supervised fine-tuning is continued training of a pretrained model on curated demonstrations — input paired with the output you want — so that the demonstrated behaviour becomes the model's default.
used by two