iterative-optimization

Tag

Cards List
#iterative-optimization

From Errors to Rules: Iterative Prompt Optimization for Text Classification

arXiv cs.AI · 2026-07-24 Cached

This paper presents ERGO, an error-driven method for iterative prompt optimization in text classification that diagnoses classification failures and generates targeted decision rules, achieving best accuracy on tasks where errors concentrate in specific confused label pairs.

0 favorites 0 likes
#iterative-optimization

Exploring Autonomous Agentic Data Engineering for Model Specialization

arXiv cs.CL · 2026-06-01 Cached

This paper formalizes Autonomous Agentic Data Engineering, where LLMs act as autonomous data engineers to curate and optimize training data for specialized domains, showing a 57.29% improvement in student model performance using GPT-5.2.

0 favorites 0 likes
#iterative-optimization

@WWTLitee: Is there a way for AI to autonomously iterate and optimize? Yes, check out autoresearch. Its core isn't to have AI directly 'invent papers,' but to break the research process into a verifiable loop: humans write program.md to give research direction, AI agent modifies http://tra…

X AI KOLs Timeline · 2026-05-23 Cached

Introduces the autoresearch project, which breaks down the AI research process into a verifiable loop (fixed environment, single editable file, fixed metric, Git rollback), enabling AI agents to perform controllable and reproducible experiment iterations; also mentions the 12-factor-agents checklist.

0 favorites 0 likes
← Back to home

Submit Feedback