Skip to content
OpenTrain AIFor AI Companies

HFEPX · Eval paper review

LEAF-OPD: Learning to Forget from Privileged Contexts via On-Policy Distillation

Zirui Pang, Zhijie Deng, Yuancheng Wang, Qihan Yu +3 more

Published

Nov 16, 2026

Citations

0

Trust level

Low

Usefulness score

25/100 (Low)

Extraction confidence

45% (Low)

Derived from extracted protocol signals and abstract evidence.

Rater population

Not reported

Signals refreshed

Nov 16, 2026

Should you rely on this paper?

This paper is adjacent to HFEPX scope and is best used for background context, not as a primary protocol reference.

Use this as background context only. Do not make protocol decisions from this page alone.

Best use

Background context only

Use if you need

A secondary eval reference to pair with stronger protocol papers.

What to verify

Validate the evaluation procedure and quality controls in the full paper before operational use.

Main weakness

The available metadata is too thin to trust this as a primary source.

Human feedback signal
Not explicit
Not explicit in abstract metadata
Evaluation signal
Detected
Eval setup described
Usefulness for eval research
25/100
Adjacent candidate

Treat as adjacent context, not a core eval-method reference.

Abstract

LLM unlearning aims to erase targeted knowledge from large language models while preserving their utility. However, many existing methods suppress target knowledge without explicitly teaching the model how to respond to queries about it. We propose LEAF-OPD (LEArning to Forget from Privileged Contexts via On-Policy Distillation), in which a frozen teacher provides token-level supervision on student-generated responses. On forget data, the teacher uses privileged contexts (instructions not to reveal knowledge about forget targets) to guide forgetting. On retain data, the teacher provides supervision without privileged context to preserve model utility. The student operates without privileged context in both cases, learning how to respond when encountering knowledge that should be forgotten. A key challenge is that the teacher often provides effective supervision only at the first few tokens of student trajectories. We introduce Supervision Signal (SS) to measure corrective supervision over a specified token interval, evaluating the entire trajectory (SS1) and its second half (SS0.5) to assess overall strength and late-stage persistence. When supervision is weak, lightweight SFT strengthens the teacher's context-conditioned refusal behavior, substantially increasing both scores. We evaluate LEAF-OPD on three widely used benchmarks: TOFU, RWKU, and MUSE-News. On TOFU, LEAF-OPD achieves the best Forget Quality, Model Utility, and Forget ROUGE-L while maintaining competitive retain performance. On RWKU, it outperforms the second-best method by 34% in Forget Quality while maintaining competitive Retain Quality. These results demonstrate the effectiveness of privileged-context supervision with on-policy token-level distillation for LLM unlearning.

What we could verify

These are the protocol signals we could actually recover from the available paper metadata. Use them to decide whether this paper is worth deeper reading.

Human Feedback Types

missing

None explicit

No explicit feedback protocol extracted.

"LLM unlearning aims to erase targeted knowledge from large language models while preserving their utility."

Evaluation Modes

partial

Automatic Metrics

Includes extracted eval setup.

"LLM unlearning aims to erase targeted knowledge from large language models while preserving their utility."

Quality Controls

missing

Not reported

No explicit QC controls found.

"LLM unlearning aims to erase targeted knowledge from large language models while preserving their utility."

Benchmarks / Datasets

missing

Not extracted

No benchmark anchors detected.

"LLM unlearning aims to erase targeted knowledge from large language models while preserving their utility."

Reported Metrics

partial

Rouge

Useful for evaluation criteria comparison.

"On TOFU, LEAF-OPD achieves the best Forget Quality, Model Utility, and Forget ROUGE-L while maintaining competitive retain performance."

Benchmarks and datasets

No benchmark or dataset names were extracted from the available abstract.

Reported metrics

rouge
Human feedback details
Uses human feedback
No
Feedback types
None
Rater population
Not reported
Unit of annotation
Trajectory (inferred)
Expertise required
General
Evaluation details
Evaluation modes
Automatic Metrics
Agentic eval
Long Horizon
Quality controls
Not reported
Evidence quality
Low
Use this page as
Background context only

Research brief

Metadata summary

LLM unlearning aims to erase targeted knowledge from large language models while preserving their utility.

Based on abstract + metadata only. Check the source paper before making high-confidence protocol decisions.

Key takeaways

  • LLM unlearning aims to erase targeted knowledge from large language models while preserving their utility.
  • However, many existing methods suppress target knowledge without explicitly teaching the model how to respond to queries about it.
  • We propose LEAF-OPD (LEArning to Forget from Privileged Contexts via On-Policy Distillation), in which a frozen teacher provides token-level supervision on student-generated responses.

Researcher actions

  • Compare this paper against nearby papers in the same arXiv category before using it for protocol decisions.
  • Check the full text for explicit evaluation design choices (raters, protocol, and metrics).
  • Use related-paper links to find stronger protocol-specific references.

Caveats

  • Generated from abstract + metadata only; no PDF parsing.
  • Signals below are heuristic and may miss details reported outside the abstract.

Recommended queries

Contribution summary

  • LLM unlearning aims to erase targeted knowledge from large language models while preserving their utility.
  • However, many existing methods suppress target knowledge without explicitly teaching the model how to respond to queries about it.

Why it matters for eval

  • We introduce Supervision Signal (SS) to measure corrective supervision over a specified token interval, evaluating the entire trajectory (SS1) and its second half (SS0.5) to assess overall strength and late-stage persistence.

Researcher checklist

  • Human feedback protocol is explicit

    No explicit human feedback protocol detected.

  • Evaluation mode is explicit

    Detected: Automatic Metrics

  • Quality control reporting appears

    No calibration/adjudication/IAA control explicitly detected.

  • Benchmark or dataset anchors are present

    No benchmark/dataset anchor extracted from abstract.

  • Metric reporting is present

    Detected: rouge