Newest papers

Loopholing All the Way: Language Modeling with pure Self-Conditioning

Adrien Deschenaux, Nadezhda Ilieva, Justin Deschenaux, Manuel Madeira, Dorina Thanou, Aurélien Lucchi, Antonio Orvieto, Caglar Gulcehre · 2026 · arXiv

DOI: 10.48550/arXiv.2610.30068 · open-access PDF

Korean

최근의 확산 언어 모델(DLMs)은 스스로 조절(SC), 또는 루프홀링을 사용하여 성능을 향상시킵니다. SC는 전통적으로 보조 진화론으로 간주되지만, 그 경험적 성공은 질문을 제기합니다: 지속적인 생성 모델이 단순히 자기 조절에 근거할 수 있습니까? 우리는 LAW(“Loopholing All the Way”)로 긍정적으로 대답합니다.이 모델은 단계별로 단계별로 예측을 통해 예상된 삽입을 전파하여 토큰을 한 번 해독합니다.

Machine translation (Meta M2M100); the English original follows.

Abstract

Recent Diffusion Language Models (DLMs) use Self-Conditioning (SC), or Loopholing, to improve their performance. While SC is traditionally seen as an auxiliary heuristic, its empirical success raises the question: can a continuous generative model be based purely on Self-Conditioning? We answer affirmatively with LAW (``Loopholing All the Way''), which propagates the Expected Embedding under the denoiser's prediction across steps, decoding the tokens once, at the end. On Sudoku, LAW outperforms previous Discrete and Continuous Diffusion Models, including Loopholing Discrete Diffusion. On Maze,

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange