cd /news/machine-learning/lost-in-interpolation-why-predictive… · home topics machine-learning article
[ARTICLE · art-89831] src=arxiv.org ↗ pub= topic=machine-learning verified=true sentiment=↑ positive

Lost in Interpolation: Why Predictive Feedback Fails in Diffusion Language Models

Researchers at arXiv introduced Spherical Soft-Masking (S-SM), a drop-in replacement for linear interpolation in Masked Diffusion Language Models (MDLMs), after finding that the embedding space is hyperspherical with a near-constant angle of approximately 73 degrees. In continued pre-training of a 169M-parameter MDLM checkpoint, S-SM with SLERP feedback avoided training degradation and delivered MAUVE gains of up to 2x over the vanilla baseline and 27.5-56.1% over TopK/LERP, with 16.9-19.6% lower generative perplexity.

read1 min views1 publishedAug 10, 2026

arXiv:2608.06529v1 Announce Type: new Abstract: Soft-masking accelerates the convergence of Masked Diffusion Language Models (MDLMs). Existing formulations build this blend with linear interpolation (LERP) in the raw embedding space, which implicitly treats that space as Euclidean. We analyze the embedding space of MDLMs and find that the mask and predicted-token embeddings maintain a near-constant angle of (\approx 73^\circ) throughout training, while embedding norms remain essentially flat across vocabulary-frequency rank. These indicate a hyperspherical geometry, for which LERP is the wrong interpolation primitive. We introduce Spherical Soft-Masking (S-SM), a drop-in replacement that aggregates the top-(k) predictions with a Fr'echet mean on the hypersphere and blends this mean with the mask direction using spherical linear interpolation (SLERP), then restores the native mask norm. We evaluate S-SM on continued pre-training of a released 169M-parameter MDLM checkpoint across a wide range of inference-time step budgets, SLERP feedback avoids the training degradation that LERP feedback induces and delivers MAUVE gains of up to 2x over the vanilla MDLM baseline and 27.5-56.1% over TopK/LERP at various sampling budgets, alongside consistently lower generative perplexity (16.9-19.6% over the baseline), while leaving output entropy and convergence essentially unchanged.

── more in #machine-learning 4 stories · sorted by recency
── more on @arxiv 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/lost-in-interpolatio…] indexed:0 read:1min 2026-08-10 ·