Jinya Sakurai, Xun Xu, Jaehong Yoon · 2026 · arXiv
DOI: 10.48550/arXiv.2610.31576 · open-access PDF
Kebijakan unmasking yang dipelajari meningkatkan kompromi kecepatan - akurasi, namun berbeda dalam pengawasan, input, dan kapasitas, membingungkan pilihan mana yang penting. Kami melihat kebijakan yang diawasi sebagai mendistribusikan prosedur decoding melalui seorang guru dan secara independen memeriksa apa yang dinyatakan oleh guru sebagai kompromi, apa yang dinyatakan oleh kebijakan itu meniru, apa yang diperhatikan oleh kebijakan, dan seberapa banyak jumlah penilaian fungsi (NFE) harus dikurangi untuk mengimbangi kebijakan.
Machine translation (Meta M2M100); the English original follows.
Masked diffusion language models generate tokens in parallel, but their efficiency depends on which predictions are committed. Learned unmasking policies improve the speed--accuracy trade-off, yet differ in supervision, inputs, and capacity, obscuring which choices matter. We view supervised policies as distilling a decoding procedure through a teacher and examine independently what the teacher labels as committable, on which states the policy imitates it, what the policy observes, and how much the number of function evaluations (NFE) must decrease to offset policy overhead. We find that label
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=2610.31576, 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