{"id":"arxiv:2604.08302","title":"DMax: Aggressive Parallel Decoding for dLLMs","date":"2026-04-09","section":"LLMs: architectures, context, training","quality":0.217022,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/","source_url":"https://arxiv.org/abs/2604.08302","text_available":true,"authors":["Zigeng Chen","Gongfan Fang","Xinyin Ma","Ruonan Yu"],"outgoing":[{"source":"arxiv:2604.08302","target":"arxiv:2407.21783","evidence":"id","id":"arxiv:2407.21783","title":"The Llama 3 Herd of Models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/de3cebb4c28bc5d41a843279f219dca14476284991ea5c7c0aeab7068a40e306/"},{"source":"arxiv:2604.08302","target":"arxiv:2502.09992","evidence":"id","id":"arxiv:2502.09992","title":"Large Language Diffusion Models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/a4863ed6f639560bd2feb6c66c13d1e3ac4cf42f8919a65c9818625cc6b6cde9/"},{"source":"arxiv:2604.08302","target":"arxiv:2512.15745","evidence":"id","id":"arxiv:2512.15745","title":"LLaDA2.0: Scaling Up Diffusion Language Models to 100B","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/90a25474fdf076c254ea3ddbbc7ddbc1f2f49fb98065168d66b5ad9d8f7dafaa/"},{"source":"arxiv:2604.08302","target":"arxiv:2602.08676","evidence":"id","id":"arxiv:2602.08676","title":"LLaDA2.1: Speeding Up Text Diffusion via Token Editing","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1b6dd6e64e69da1fff43a183af57a467878db80d54f550bceefa049d7bb18878/"}],"incoming":[],"text_parts":[{"start":0,"end":15479,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-1.html"},{"start":15479,"end":31061,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-2.html"},{"start":31061,"end":31630,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-3.html"}],"contents":[{"title":"DMax: Aggressive Parallel Decoding for dLLMs","offset":0,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-1.html#text"},{"title":"Abstract","offset":48,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-1.html#text"},{"title":"1 Introduction","offset":1392,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-1.html#text"},{"title":"2 Preliminaries","offset":5374,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-1.html#text"},{"title":"3 Methodology","offset":8840,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-1.html#text"},{"title":"3.1 On-Policy Uniform Training","offset":8858,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-1.html#text"},{"title":"3.2 Soft Parallel Decoding","offset":14128,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-1.html#text"},{"title":"4 Experiments","offset":19843,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-2.html#text"},{"title":"4.1 Experimental Setup","offset":19861,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-2.html#text"},{"title":"4.2 Experimental Results","offset":23417,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-2.html#text"},{"title":"5 Ablation Study","offset":26358,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-2.html#text"},{"title":"6 Related Work","offset":28605,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-2.html#text"},{"title":"7 Conclusion","offset":31044,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/part-2.html#text"}],"text_code_points":31630,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/22790263bd218aac28626fe3dab2fcd0ed73ceef94aa8a948bef2cb8bc007f7a/source.md"}