arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2606.30705cs.LGcs.AI
收藏

为什么少步文本潜在变量在图像潜在变量有效时失败?尖锐分类读出中的非承诺性

Why Do Few-Step Text Latents Fail When Image Latents Work? Non-Commitment at Sharp Categorical Readouts

Fudan University(复旦大学);Shanghai Innovation Institute(上海创新研究院)

Zhongyao Wang

AI总结:

本文发现确定性少步生成在连续图像潜在变量上成功,但在连续文本潜在变量上崩溃为不连贯文本,原因是几何因素而非训练或缩放缺陷:平滑的确定性映射无法在尖锐分类读出前解决离散分支选择,失败由解码器尖锐度而非传输精度主导。

摘要:

确定性少步生成在连续图像潜在变量上成功,但在连续文本潜在变量上崩溃为不连贯文本,我们证明原因是几何因素而非训练或缩放缺陷:平滑的、受正则性限制的确定性映射无法在尖锐分类读出前解决离散分支选择,因此少步失败由解码器尖锐度而非传输精度主导。在真实文本自编码器的重叠区域,我们证明(定理3)后验均值终端步骤以决策边界周围$O(s(t))$管内的潜在质量速率翻转令牌。两个诊断指标DABI(读出尖锐度)和CCI(分类承诺)在已发布检查点上测量表明,四个独立构建的连续文本解码器将边界对齐扰动放大远超范数匹配的各向同性扰动(DABI从$5\times10^{2}$到$>10^{5}$),而图像解码器的DABI约为1。两种机制规避了连续界限:分类承诺(自回归解码器尽管读出更尖锐仍成功)和随机再注入(同一模型上确定性ODE在$K=4$时PPL为294,而SDE为50)。在理想化的分离区域,我们证明了匹配的尖锐传输定律,包括维度相图:分离$M$模式所需的确定性刚度在潜在维度为$\Omega(\log M)$时增长为$\Theta(\sqrt{\log M})$(在固定维度中为$M^{1/n}$),深度为$B$的层次结构给出每步峰值小$\sqrt{B}$(定理5-7);一个余面积恒等式将这些与重叠管联系起来(定理17)。结果是精度-深度-刚度权衡:在确定性-连续类中,代价不可约,两种规避都走出了该类。

英文摘要:

Deterministic few-step generation succeeds on continuous image latents but collapses to incoherent text on continuous text latents, and we show the cause is geometric rather than a training or scaling deficiency: a smooth, regularity-limited deterministic map cannot resolve a discrete branch choice before a sharp categorical readout, so few-step failure is governed by decoder sharpness, not transport accuracy. In the overlapping regime of real text autoencoders, we prove (Theorem 3) that the posterior-mean terminal step flips tokens at the rate of the latent mass in an $O(s(t))$ tube around decision boundaries. Two diagnostics, DABI (readout sharpness) and CCI (categorical commitment), measured on published checkpoints show that four independently built continuous-text decoders amplify a boundary-aligned perturbation far beyond a norm-matched isotropic one (DABI from $5\times10^{2}$ to $>10^{5}$), while image decoders have DABI $\approx 1$. Two mechanisms escape the continuous bound: categorical commitment (autoregressive decoders succeed despite sharper readouts) and stochastic re-injection (deterministic ODE at $K=4$ gives PPL 294 versus SDE 50 on the same model). In the idealized separated regime we prove matching sharp transport laws, including a dimension phase diagram: the deterministic stiffness needed to separate $M$ modes grows as $Θ(\sqrt{\log M})$ once the latent dimension is $Ω(\log M)$ (and as $M^{1/n}$ in fixed dimension), with a depth-$B$ hierarchy giving a $\sqrt{B}$-smaller per-step peak (Theorems 5-7); a coarea identity links these to the overlapping tube (Theorem 17). The result is an accuracy-depth-stiffness tradeoff: within the deterministic-continuous class the cost is irreducible, and both escapes step outside it.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑