AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

Generative Diffusion Prior Distillation for Long-Context Knowledge Transfer

Hewa Dehigahawattage Nilushika Udayangani, Nandakishor Desai, Marimuthu Palaniswami

ICLR 2026 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train diffusion prior with warm-up phase (E_warm=300) instead of none (ablation)GDPD · StarLightCurves e=0.5LAUC-PRC: 85.56 -> 97.26Improved
jointly train diffusion prior with GDPD loss without warm-up (ablation)GDPD · StarLightCurves e=0.5LAUC-PRC: 85.56 -> 89.72Improved
train the diffusion prior at all (ablation)GDPD · StarLightCurves e=0.5LAUC-PRC: 83.12 -> 97.26Improved
combine task loss with GDPD loss instead of GDPD only (ablation)GDPD · StarLightCurves e=0.5LAUC-PRC: 85.72 -> 97.26Improved
use GDPD loss instead of logit-KD loss (ablation)GDPD · StarLightCurves e=0.5LAUC-PRC: 94.51 -> 97.26Improved
use GDPD loss instead of feature-KD loss (ablation)GDPD · StarLightCurves e=0.5LAUC-PRC: 83.81 -> 97.26Improved
use more posterior samples J=4 instead of J=1 (ablation)GDPD · StarLightCurves e=0.5LAUC-PRC: 97.26 -> 97.87No clear effect
use GDPD instead of Base-KD for partial classificationLSTM3-100 · 12 UCR e=0.5Lavg AUC-PRC: 78.14 -> 84.64Improved
use GDPD instead of feature KD (Fits) for compression to LSTM1-8LSTM3-100 · 12 UCR e=0.5Lavg AUC-PRC: 59.84 -> 72.76Improved
use GDPD for self-distillation instead of Base-KDLSTM3-100 · 12 UCR e=Lavg AUC-PRC: 89.29 -> 91.21Improved
use GDPD for self-distillation instead of Base-KDResNet32-64 · 12 UCR e=Lavg AUC-PRC: 98.52 -> 98.58No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy