{"slug": "agentic-on-policy-distillation-implicitly-reshapes-llms-react-patterns", "title": "Agentic On-Policy Distillation Implicitly Reshapes LLMs' ReAct Patterns", "summary": "On-policy distillation (OPD) implicitly reshapes the ReAct interaction patterns of language-model agents across mathematics, ALFWorld and τ^2-bench, according to a study examining OPD's cross-domain effects on reasoning, actions and observations. The research reports that OPD transfers specialized capabilities to LLM agents but that its effects on interaction patterns beyond the target domain remain unclear.", "body_md": "On-policy distillation (OPD) transfers specialized capabilities to language-model agents, but its cross-domain effects on interaction remain unclear. We investigate how OPD reshapes ReAct interaction patterns across mathematics, ALFWorld, and τ^2-bench, focusing on reasoning, actions, and observatio", "url": "https://wpnews.pro/news/agentic-on-policy-distillation-implicitly-reshapes-llms-react-patterns", "canonical_source": "https://aiflash.com/news/153544/", "published_at": "2026-11-18 05:02:14+00:00", "updated_at": "2026-11-18 05:16:36.822539+00:00", "lang": "en", "topics": ["ai-research", "large-language-models", "ai-agents", "machine-learning"], "entities": ["ALFWorld", "τ^2-bench", "ReAct"], "also_reported_by": [], "takeaway": "On-policy distillation changes an LLM agent's ReAct reasoning, action and observation patterns beyond the domain it was trained on, so cross-domain interaction behavior should be measured on benchmarks like ALFWorld and τ^2-bench rather than assumed stable.", "recommendation": {"advice": "After on-policy distillation, re-run ReAct interaction evals on out-of-domain benchmarks such as ALFWorld and τ^2-bench to catch shifted reasoning, action and observation patterns before deployment.", "playbook": "evals-testing", "playbook_url": "https://wpnews.pro/playbooks/evals-testing"}, "alternates": {"html": "https://wpnews.pro/news/agentic-on-policy-distillation-implicitly-reshapes-llms-react-patterns", "markdown": "https://wpnews.pro/news/agentic-on-policy-distillation-implicitly-reshapes-llms-react-patterns.md", "text": "https://wpnews.pro/news/agentic-on-policy-distillation-implicitly-reshapes-llms-react-patterns.txt", "jsonld": "https://wpnews.pro/news/agentic-on-policy-distillation-implicitly-reshapes-llms-react-patterns.jsonld"}}