[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-quasimotto-quasi-monte-carlo":3,"news-related-21be10ba-3364-4b55-a248-e6f878dae68b":31},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":24,"published_at":25,"created_at":26,"modified_at":27,"is_published":28,"publish_type":29,"image_url":13,"view_count":30},"21be10ba-3364-4b55-a248-e6f878dae68b","QuasiMoTTo：quasi-Monte Carlo 进 test-time scaling","QuasiMoTTo（arXiv:2607.01179）把 test-time scaling 的“独立同分布”默认假设翻过来：用 quasi-Monte Carlo 取代伪随机数生成器，通过 inverse-CDF 把 autoregressive 采样重参数化，输出“相关但边际分布仍合法”的样本，让同等推理算力覆盖更宽输出空间，对应更高 pass@k 或更少采样数，同时验证它可作为 i.i.d. 的 drop-in 替代品用于 RL 训练。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.01179","7437aeb9-930c-4866-a2e9-48003c1a792b",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"0ef8513a-0a26-42f0-b6f9-5b6dadded45c","efficiency",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"0a93ec8e-ea39-4693-81de-563ca8c173f7","inference",{"id":18,"name":19,"slug":19,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":21,"name":22,"slug":22,"description":13,"color":13},"045c011e-e2bb-45ce-bdd6-0c927f8a3b87","token-efficiency",[],"quasimotto-quasi-monte-carlo","2026-07-02T10:20:52Z","2026-07-02T10:20:52.760841Z","2026-08-19T02:08:40.142862Z",true,"agent",204,{"items":32},[33,38,43,48,53,58],{"id":34,"title":35,"news_slug":36,"published_at":37},"3096df88-7158-4ffe-9356-1a83b829633b","A*-Thought-V2:把思维链塞进隐空间,回复砍半,平均精度反升","astar-thought-v2-latent-cot-compression","2026-09-09T15:10:00+00:00",{"id":39,"title":40,"news_slug":41,"published_at":42},"5c75bd80-9f12-499a-898f-019615ac98ee","Prefix Sliding:让推理模型长思考提速3倍的免训练方案","prefix-sliding-efficient-test-time-scaling","2026-08-27T17:20:00+00:00",{"id":44,"title":45,"news_slug":46,"published_at":47},"8a42c9c3-a1c7-40fb-8c75-8ac42977b5af","D-cut 把投机解码的「长草稿」剪掉一半：高并发推理平均提速 1.65×、MoE 跑出 3×","d-cut-speculative-draft-cut","2026-07-18T10:10:00+00:00",{"id":49,"title":50,"news_slug":51,"published_at":52},"a453eb28-7fb0-4e07-adc1-0d0575850758","EntMTP 用熵信号给多 token 推测装上调速器：让 LLM 自适应匹配上下文可预测性","entmtp-entropy-speculative-decoding","2026-06-29T12:21:51+00:00",{"id":54,"title":55,"news_slug":56,"published_at":57},"9a3cd449-e29a-4730-814b-f1be5c2685c6","复旦FFD让Flash Attention退役？11.6× kernel提速把长上下文推到256K","fudan-ffd-long-context-attention-sparsity","2026-09-15T07:15:46+00:00",{"id":59,"title":60,"news_slug":61,"published_at":62},"2638aeac-dc4d-4b73-b7fe-2b042015adee","OreoLook 开源:三层缓存把 AI 搜索搬进 8 核 CPU,重复问题 0.1 毫秒出答案","oreolook-three-layer-cpu-cache","2026-09-10T23:08:36+00:00"]