Trendoday

Restoring Diversity in Distilled Image and Video Generators by Probing Activations

Distilling diffusion models into few-step student models substantially reduces sampling latency, but induces severe diversity collapse: samples generated from different random seeds show minimal varia

  • Upvotes4
–Rank now
#23Peak rank
2Days on list
43Score · 7d
Nov 17, 2026First seen
Rank · 7 days

About this paper

Distilled few-step diffusion models exhibit diversity collapse because activation variance is concentrated into fewer directions in an intermediate block at the initial denoising step, and ReDiverse restores sample diversity by reintroducing teacher variance directions through adapter fine-tuning…

Distilling diffusion models into few-step student models substantially reduces sampling latency, but induces severe diversity collapse: samples generated from different random seeds show minimal variation. In this work, we investigate the root cause of this failure by probing the internal representations of student models relative to their teachers. We find that distilled students maintain total activation variance comparable to their teachers, but concentrate it into substantially fewer directions. This directional collapse is localized to an identifiable intermediate block at the initial denoising step. Consequently, diversity collapse stems from a misallocation of activation variance…

diffusion modelsfew-step student modelsdiffusion model distillationdiversity collapseinternal representationsactivation variancedirectional collapseintermediate blockinitial denoising stepReDiverse
Upvotes
4
Authors
7
Published
Nov 13, 2026

Summary and abstract excerpt via Hugging Face Papers ↗ · arXiv.

Trending history

Restoring Diversity in Distilled Image and Video Generators by Probing Activations was first seen on Hugging Face · Daily papers on Nov 17, 2026. Since then it has appeared in 7 checks on 2 different days. Its best position was #23 of 30, reached on Nov 17, 23:28 UTC. It dropped off the list after Nov 18, 01:28 UTC.

Rank history

Position on Hugging Face · Daily papers at every check (lower is better).

#1#11#20#30 13:2816:2819:2822:2801:28 Nov 17, 13:28 UTC · #29Nov 17, 15:28 UTC · #26Nov 17, 17:28 UTC · #28Nov 17, 19:28 UTC · #30Nov 17, 21:28 UTC · #27Nov 17, 23:28 UTC · #23Nov 18, 01:28 UTC · #23

Last 14 days

Best rank per day (UTC).

Nov 9today

Metric history

Platform metric as reported by Hugging Face.

0235 13:2816:2819:2822:2801:28 Nov 17, 13:28 UTC · 4Nov 17, 15:28 UTC · 4Nov 17, 17:28 UTC · 4Nov 17, 19:28 UTC · 4Nov 17, 21:28 UTC · 4Nov 17, 23:28 UTC · 5Nov 18, 01:28 UTC · 5

Questions

Is Restoring Diversity in Distilled Image and Video Generators by Probing Activations trending on HF Daily Papers right now?

Not at the moment. It was last on the list on Nov 18, 01:28 UTC, after 2 days of trending.

What is the highest rank Restoring Diversity in Distilled Image and Video Generators by Probing Activations reached?

#23 on Hugging Face · Daily papers, on Nov 17, 23:28 UTC. Trendoday checks this list every 120 minutes.

How long has Restoring Diversity in Distilled Image and Video Generators by Probing Activations been trending?

It was first seen on Nov 17, 2026 and has been on the list on 2 days (7 checks) since.