简体中文 · English

ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents

L1 core mechanism diagnostic. The local implementation runs two levels of state updates, with no actual lab connection.

Paper information

Field	Details
Paper	ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents
Institution	Gen-Verse research collaboration (first author affiliation)
First public release	2026-09-15 (arXiv v1)
Original source code	Yes: Gen-Verse/ScienceBuddy
Adapter/method	sciencebuddy
Local reproduction code	src/auto_research/agent_research/latest_20260916.py

Original paper summary

Background and main changes

The inner recursion evolves the interaction harness. The outer recursion updates the scientific agent's policy from successful trajectories.

The local reference kernel retains the defining gates, weights or state transitions and emits auditable intermediates for comparison under a shared budget.

flowchart LR
  I[Public inputs and state] --> M[ScienceBuddy core mechanism]
  M --> A[Auditable intermediates]
  A --> O[Output or policy update]

Key figure from the paper

Figure 4: Core architecture, modules and connections. From the paper; copyright belongs to the authors. Click for the source.

Core mechanism

The implementation preserves the defining state transitions and exposes relevant weight, gate, depth, stage or cache statistics. Fixed budgets and three seeds check mechanism invariants, not paper-scale training.

Offline and online results

Capability, speed and cost claims are attributed to the paper only. Local results are not directly compared or extrapolated to online gains.

Local reproduction

See metrics/mechanism-seeds42-44.json. The unified receipt marks diagnostic_only=true, so the public dashboard does not treat it as a full capability gain.

Scope

Two levels of state updates run locally. There is no connection to an actual lab environment.