Mahdi Sabbaghi, George Pappas, Hamed Hassani, Adel Javanmard · 2026 · arXiv
DOI: 10.48550/arXiv.2610.31563 · open-access PDF
Distilasi Kebijakan Online (OPD) sering melampaui metode offline seperti Supervised Fine-Tuning (SFT) dengan pelatihan pada trajektori yang diambil sampel langsung dari siswa. Namun, OPD standar melatih siswa untuk mengikuti guru di setiap kondisi trajektori, dengan asumsi bahwa bimbingan guru selalu menguntungkan. Sebaliknya, kami membedakan antara apa yang disukai guru dan apa yang benar-benar membantu siswa saat ini berhasil. Kami memperkenalkan Distilasi Berbasis Keuntungan pada Kebijakan (AOPD), yang menimbang tindakan guru dengan keuntungan kontrafaktual mereka di bawah siswa.
Machine translation (Meta M2M100); the English original follows.
Online Policy Distillation (OPD) often outperforms offline methods such as Supervised Fine-Tuning (SFT) by training on trajectories sampled directly from the student. However, standard OPD trains the student to follow the teacher at every state of the trajectory, assuming that teacher guidance is always beneficial. We instead distinguish between what the teacher prefers and what actually helps the current student succeed. We introduce Advantage-Based On-Policy Distillation (AOPD), which weights teacher actions by their counterfactual advantage under the student. Teacher actions that improve th
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.31563, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange