recursive

18 stories and discussions about recursive, aggregated from every source we track.

1.

Ramez Naam gives a skeptic’s take on Recursive Self-Improvement.

4 points•anon373839•about 2 months ago•0 comments•
2.

Ramez Naam gives a skeptic’s take on Recursive Self-Improvement.

4 points•pretext•about 2 months ago•0 comments•
3.

Contribute to google-research/rrsi development by creating an account on GitHub.

3 points•simonpure•about 2 months ago•0 comments•
4.

I'm already exhausted of hearing about Recursive Self Improvement

2 points•xondono•about 2 months ago•0 comments•
5.

Resilient parsing means that we try to parse as much as possible of the source code, possiblyproducing multiple parsing errors. Clearly, this is useful in numerous IDE features likeauto-completion, and can be a nice…

2 points•matt_d•about 2 months ago•0 comments•
6.

Regularize the search, not the harness. Harness evolution that transfers out of distribution on a third fewer policy tokens.

2 points•jonbaer•about 2 months ago•0 comments•
7.

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…

2 points•simonpure•about 2 months ago•0 comments•
8.

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…

2 points•handfuloflight•about 2 months ago•1 comment•
9.
2 points•nicbogaert•2 months ago•0 comments•
10.

A lesson from Turing on how mathematics can make difficult questions answerable

1 points•martintrif•about 2 months ago•0 comments•
11.

Regularize the search, not the harness. Harness evolution that transfers out of distribution on a third fewer policy tokens.

1 points•yarapavan•about 2 months ago•1 comment•
12.

Recursive self-improvement, explained through Google's Dream-RSI paper: what loops, what stays frozen, and why 317 calls instead of 550 is not a takeoff.

1 points•axrisi•about 2 months ago•0 comments
13.

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…

1 points•sbulaev•2 months ago•0 comments•
14.
1 points•paulpauper•2 months ago•0 comments•
15.

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…

1 points•Betelbuddy•2 months ago•0 comments•
16.
1 points•helloplanets•2 months ago•0 comments•
17.

A visual Python coordinator for Claude Code and Codex desktop apps, with shared tasks, review checkpoints, and automated checks. - sealvlon/Recursive-Cognitive-Optimization-RCO-

1 points•sealvarezl•2 months ago•0 comments•
18.

“We never want to be in a situation again where we underestimate the AI.”

1 points•gmays•2 months ago•0 comments•

Related topics