distillation

RSS

1
When EOS Tokens Disagree: Understanding Length Inflation in On-Policy Distillation arxiv.org
On-policy distillation (OPD) can lead to excessively long student responses, a phenomenon called length inflation. This paper identifies "termination-token mismatch" between base students and post-trained teachers as a key source, where models place stopping probability on different EOS tokens. Treating functionally equivalent EOS tokens as a shared semantic stopping action substantially mitigate…
Hugging Face Daily Papersarxiv.orgYuxiao Yang, Tianrun Yu, Shangzhe Li5d1 minpaper
2
RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning arxiv.org
RetireOPD introduces a self‑retiring on‑policy distillation framework for multi‑turn RL agents. A skill‑conditioned teacher is first trained with environment rewards, then a skill‑free student learns jointly via RL and token‑level distillation. The student automatically drops the teacher once its performance gap stops shrinking and it reaches a target success‑rate fraction, after which training c…
Hugging Face Daily Papersarxiv.orgYan Yu, Zhengxi Lu, Yizhou Liu5d1 minpaper
3
Training Specialist Models without Reasoning Trajectories for Domain Expert Distillation arxiv.org
The paper shows that when specialist LLMs are trained only on QA pairs (no explicit reasoning supervision), their optimization implicitly selects a latent distribution of reasoning trajectories. By treating the distilled student as an agnostic probe—since it inherits only the sampled trajectories—the authors empirically demonstrate a strong correlation (across 27 specialist‑student pairs) between…
Hugging Face Daily Papersarxiv.orgYilei Tu, Zihao Li, Shaoxiong Ji10d1 minpaper
4
Trajectory as the Teacher: Few-Step Discrete Flow Matching via Energy-Navigated Distillation apple.com
Apple researchers propose Trajectory‑Shaped Discrete Flow Matching (TS‑DFM), a training‑time distillation method that replaces blind stochastic jumps in discrete flow‑matching with an energy‑based compass to select higher‑quality intermediate tokens. On a 170 M‑parameter language model, the 8‑step student outperforms the 1 024‑step teacher by 32 % perplexity while being 128× faster, beating basel…
Apple Machine Learning Researchapple.com6d1 minpaper
5
Lightning Weave: Improving the Accuracy-Efficiency Frontier of Reasoning Models through Capability Composition arxiv.org
Lightning Weave is a post‑training framework that composes specialist capabilities from separate models into one student via on‑policy distillation, improving both accuracy and token efficiency on math and code benchmarks.
Hugging Face Daily Papersarxiv.orgYecheng Wu, Song Han, Han Cai9d1 minpaper

distillation

When EOS Tokens Disagree: Understanding Length Inflation in On-Policy Distillationarxiv.org

RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learningarxiv.org

Training Specialist Models without Reasoning Trajectories for Domain Expert Distillationarxiv.org

Trajectory as the Teacher: Few-Step Discrete Flow Matching via Energy-Navigated Distillationapple.com

Lightning Weave: Improving the Accuracy-Efficiency Frontier of Reasoning Models through Capability Compositionarxiv.org

When EOS Tokens Disagree: Understanding Length Inflation in On-Policy Distillation arxiv.org

RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning arxiv.org

Training Specialist Models without Reasoning Trajectories for Domain Expert Distillation arxiv.org

Trajectory as the Teacher: Few-Step Discrete Flow Matching via Energy-Navigated Distillation apple.com

Lightning Weave: Improving the Accuracy-Efficiency Frontier of Reasoning Models through Capability Composition arxiv.org