Skill Self-Play: Pushing the Frontier of LLM Capability with Co-Evolving Skills

Key points

• It uses reusable skills as the main carrier so that task diversity and verification reliability are both preserved.
• A proposer, a solver, and a skill controller all evolve together in the reinforcement-learning loop.
• The skill library creates about 20 new skills per iteration, updates existing skills, and drops useless ones.
• With Qwen3-4B, it gains up to 42.9 points on tool use, BFCL, and 12.0 points on logical reasoning, ZebraLogic.
• Dynamic skill evolution is the key, and it adds another 2.6 points in overall accuracy compared with self-learning without skills.
• Even models that start out misaligned show a large turnaround effect.

Paper links

External research summaries. These are not HDATF publications or measured product results.