ShortOPD: Recovering Pruned LLMs with Short-to-Long On-Policy Distillation
Structured pruning is a hardware-friendly way to compress LLMs, but it is mostly validated on multiple-choice recognition tasks, while the same compressed checkpoints can collapse on the free-form generation that deployment actually requires. Two observations trace this gap. First, greedy pass@1 nearly vanishes after compression, yet pass@k recovers substantially under repeated sampling: useful generations are demoted, not erased. Second, the recoverable regime fails mainly through suffix repeti
Record details
Published: 13 July 2026
Source: HuggingFace Daily Papers
Category: Research
Topics: Regulation
Retrieved: 16 July 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
Operationalising Multi-Dimensional Evaluation for Conversational Agents: A Scalable, Governed Pipeline with Selective Re-evaluation and Model Benchmarking
arXiv · 13 July 2026
It is not enough to give your moderation rules to ChatGPT: Policy-as-Prompt Moderation and Its Potential Impacts on Community Governance
arXiv · 13 July 2026
A Minimalist Retargeting-Guided Reinforcement Learning Recipe for Dexterous Manipulation
arXiv · 13 July 2026
Cost-Governed RAG: Unified Per-Tenant Cost Attribution Across Retrieval and Generation in Multi-Tenant LLM Systems
arXiv · 13 July 2026
Securing LLMs in the Wild: Privacy and Security Challenges at the Edge
arXiv red teaming query · 13 July 2026
Deborah Hughes Hallett
Harvard Kennedy School · 13 July 2026
How to cite this record
ethics.ai (13 July 2026), “ShortOPD: Recovering Pruned LLMs with Short-to-Long On-Policy Distillation,” evidence record 10560, https://ethics.ai/record/10560 (originally published by HuggingFace Daily Papers).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.