$DSTL distillation

Latest papers

1. Latent Compaction with Test-Time Training cs.CL · 14 Oct · 1 hit

   …reconstruction-selected entries held fixed. It then jointly optimizes the entire cache through end-to-end distillation from the full-cache model. Model weights remain frozen throughout, and training uses only…

2. Recovering Reliable Supervision via Critique-Guided On-Policy Distillation cs.LG · 14 Oct · 2 hits

   On-policy distillation (OPD) trains a student model on its own rollouts with token-level supervision from a teacher,…

3. VASPO: Self-Distillation with Privileged Value Estimation cs.CL · 14 Oct · 3 hits

   On-policy self-distillation provides dense supervision for reasoning models via a privileged self-teacher. However, this…

4. Reynolds-number-dependent entrance bias in temperature polarization of lab-scale membrane distillation physics.flu-dyn · 14 Oct · 3 hits

   Lab-scale membrane distillation (MD) modules are commonly used for membrane characterization and developing empirical…

5. Harness Policy Distillation: Aligning Agent Policies with Runtime Semantics cs.LG · 14 Oct · 2 hits

   …action was taken but leaves the runtime semantics behind that action implicit. We introduce Harness Policy Distillation (HPD), which makes these runtime control semantics an explicit training target alongside actions…

6. Glance Forcing: Making Bidirectional Diffusion Autoregressive at a Glance cs.CV · 14 Oct · 3 hits

   …causing severe representation interference. Motivated by this, we propose Glance Forcing, a lightweight distillation framework that decouples the denoising trajectory into two complementary slow and fast experts,…

7. Radar-Only Long-Term Beam Prediction in mmWave Vehicular Networks: An Empirical Study of Doppler Input and Cross-Modal Distillation eess.SP · 14 Oct · 4 hits

   …base station, although holding the last measured beam beats both predictors there. Cross-modal knowledge distillation from a motion-mask vision-radar teacher, by contrast, changes received power by -0.01 dB…

8. RM-OPD: On-Policy Distillation for Generative Reward Models cs.LG · 14 Oct · 5 hits

   …rise to imbalanced capability acquisition and pervasive reward hacking. Inspired by the success of On-Policy Distillation (OPD) in multi-domain capability integration, we introduce RM-OPD (Reward Modeling via On-Policy…