SOURCE-LINKED INTELLIGENCE

Extremely Sparse Supervision Incentivizes Reasoning Ability

arXiv · AI, language, vision and robotics · article · Sep 3, 2026 · UTC

Large language models demonstrate increasingly strong reasoning capabilities through effective post-training. Yet, prevailing post-training methods optimize over massive numbers of tokens, implicitly assuming that effective learning must be token-intensive. We revisit this assumption in the on-policy distillation (OPD) setting, which naturally admits dense teacher supervision at every generated token. Using the Qwen3 family, we discover a counter-intuitive phenomenon: reasoning can be effectively incentivized by an extremely small fraction of generated tokens--as few as one or two tokens per r

Read original source ↗ Open in workspace

recordType

paper

region

Global