improvement
12 stories and discussions about improvement, aggregated from every source we track.
Contribute to google-research/rrsi development by creating an account on GitHub.
Regularize the search, not the harness. Harness evolution that transfers out of distribution on a third fewer policy tokens.
On-policy distillation (OPD) trains a student model by having it generate trajectories, then matching its next-token predictions with an external teacher's next-token predictions. This provides dense, token-level…
AI agents are beginning to automate research and development across the AI stack, from improving training efficiency to optimizing inference. A natural next step is to improve the research efficiency of the agents…
AI self-improvement may not be powerful enough to overcome diminishing returns. Even RSI may not lead to a fast takeoff to ASI (superintelligence).
AI agents are beginning to automate research and development across the AI stack, from improving training efficiency to optimizing inference. A natural next step is to improve the research efficiency of the agents…
An LLM agent's capability is largely magnified by its harness, namely the prompts, control flow, tooling, memory, and context management surrounding the frozen backbone model. Recent methods increasingly automate this…
“We never want to be in a situation again where we underestimate the AI.”