skillopt

Tag

Cards List
#skillopt

@FinanceYF5: 2/ SkillOpt: Treating Documents as Trainable Parameters Microsoft treats SKILL.md as trainable model parameters—without changing weights, only optimizing natural language documents, with a validation gate filtering each change. 6 Benchmarks, 52 consecutive wins, GPT-5.5 conversation boost…

X AI KOLs Following · 3d ago Cached

Microsoft proposes the SkillOpt method, which treats documents as trainable parameters. By optimizing natural language documents without modifying weights, it improves model performance. It achieves 52 consecutive wins across 6 benchmarks, with GPT-5.5 improving by 23.5 points and Claude Code by 19.1 points.

0 favorites 0 likes
#skillopt

@Xudong07452910: Previously introduced SkillOpt, which is more about: how to repeatedly polish a skill into a more reliable "job description". The focus of this MUSE-Autoskill paper is different; it concerns how an Agent manages an entire skill library. The paper describes sk...

X AI KOLs Timeline · 6d ago Cached

This MUSE-Autoskill paper focuses on how an Agent manages an entire skill library, placing skills into a complete lifecycle: creation, memory, management, evaluation, and re-optimization.

0 favorites 0 likes
#skillopt

@DAIEvolutionHub: MICROSOFT JUST OPEN-SOURCED A WAY TO “TRAIN” AI AGENTS WITHOUT TOUCHING MODEL WEIGHTS SkillOpt treats a simple markdown…

X AI KOLs Timeline · 2026-05-28 Cached

Microsoft open-sourced SkillOpt, a method that treats markdown skill files like neural network parameters to train AI agents without modifying model weights, using learning rates, validation checks, minibatches, and epochs.

0 favorites 0 likes
#skillopt

@daniel_mac8: Codex Pro tip: turn Codex into a research engineer. Take any new agent paper and: 1. Ask: “How would this work in Codex…

X AI KOLs Timeline · 2026-05-26 Cached

A tip for Codex users to implement agent research papers directly into their Codex environment using goal mode and local config, with SkillOpt as an example that improved a GPT-5.5 agent by +24.8 points.

0 favorites 0 likes
← Back to home

Submit Feedback