Different Feedback, Different Updates: Selective Self-Learning from User Interactions for Large Language Models
2026-08-10 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors present SLIFT, a method that improves large language models (LLMs) using detailed user feedback. They break down feedback into three types: necessary fixes (Fix), optional task improvements (Spec), and irrelevant comments (Null). SLIFT trains two small adapters—a Generalist for core fixes and a Specialist for specific refinements—on top of a frozen main model. This helps the model learn the right changes at the right scale without confusing unrelated feedback. Their experiments show SLIFT works well on benchmark tests, and they have shared their code publicly.
Large Language ModelsUser FeedbackSelf-LearningLoRA AdaptersSelf-DistillationTask RefinementBehavioral ChangeModel Fine-TuningMemoryBenchWildFB
Authors
Xuanchen Li, Haitao Li, Yujia Zhou, Qingyi Pan, Heng Wang, Yiqun Liu, Min Zhang, Qingyao Ai
Abstract
User feedback offers natural supervision for persistent LLM improvement, but a single message may support multiple behavioral changes with different scopes of generalization. We introduce SLIFT, a selective self-learning framework built on a task-relative view of user feedback. SLIFT decomposes each feedback message into atomic components and interprets each component relative to the original task as Fix, Spec, or Null: requirements for task validity, compatible condition-specific refinements, or content with no reliable positive update direction. To incorporate each change at the appropriate scope, SLIFT trains two complementary LoRA adapters on a shared frozen backbone: a Generalist that consolidates Fix requirements into default behavior through feedback-conditioned self-distillation, and a Specialist that observes only the task and Generalist response to supply residual guidance for applicable, unmet Spec refinements. Null components induce no positive update. Across backbones, SLIFT achieves strong performance on both MemoryBench and WildFB, with targeted analyses further examining its underlying mechanisms. We release our code at https://anonymous.4open.science/r/SLIFT.