Adversarial Speculative Reasoning Improves Long Horizon Agents
Training draft models to reject at critical decision points improves agent accuracy by training 0.06% to 0.28% of parameters on agentic coding benchmarks.
Stories per day · 30 days
Relatedlong-horizon-agents2speculative-decoding2adversarial-preference1adversarial-reasoning1agentic-reasoning1
2 stories, newest first · one card per story
Stories tagged #draft-models
Adversarial speculative decoding improves long-horizon agent reasoning
Training draft models to be rejected at critical points raises end-to-end success on complex agentic coding benchmarks without changing target model output.