Learning User Preferences Through Interaction for Long-Term Collaboration

Open paper

Benchmark	Model	Rank	Results
mathematical-reasoning-on-math-500	gpt-oss-20b + memory	#112	Accuracy: 70.35