NVIDIA's PivotOPD Teaches Agents to Prevent and Recover From Pivotal Mistakes
nvidia · hf · 2026-10-01
PivotOPD is an on-policy distillation framework that trains agents both to avoid pivotal mistakes (found in over half of failed rollouts across Qwen3 8B-235B, usually early) and to recover from the states they create. It beats 13 baselines, gains +5.5% on ALFWorld with the 1.7B student, and lifts a Nemotron-3.5 student's SWE-Bench Verified resolve rate by 3.2%.
More from coding & agent
- Generic API-catalog MCP server vs one tool per job: a context-bloat experiment — MandeepKaur9407 · 2026-11-16
- Berkeley paper: LLMs answer with stale values even when the update is in context — ombre-ai · 2026-11-16
- Effective HTML: 4.2k-star agent skills turn wireframes into self-contained interactive prototypes — tom_doerr · 2026-11-16
- skills-gate: one gate skill fronts 148 agent skills at just 389 tokens — Distinct-Pie2389 · 2026-11-16
- Open-weight Karpo-Decision-31B tops Decision Index at 73.9, one forward pass per answer — CodeByPoonam · 2026-11-16
- shadcn boosts @uisdk: one-shot UI rebuild that's instantly E2E-testable by AI — shadcn · 2026-11-16