llm-kb
← Back to research
Paper

Researchers identify asymmetric generalization problem in LLM unlearning benchmarks

A new arXiv paper argues that existing machine unlearning benchmarks for LLMs suffer from under-forgetting and over-forgetting due to an asymmetric generalization problem. The authors propose that evaluation must cover diverse query formulations of target facts to reliably measure knowledge removal while preserving unrelated capabilities.

0 engagement·1 source·Fri, Jul 10, 2026, 09:31 AM
The paper, titled 'Forget Narrowly, Retain Broadly: Unlearning as an Asymmetric Generalization Problem,' was published on arXiv on July 10, 2026. It critiques current unlearning benchmarks for failing to detect knowledge that resurfaces under paraphrased or indirect queries (under-forgetting) and lacking semantic, syntactic, and lexical probes to verify preservation of unrelated knowledge (over-forgetting). The authors frame unlearning as an asymmetric generalization problem and call for more comprehensive evaluation methods.

Entities

arXiv(company)Forget Narrowly, Retain Broadly: Unlearning as an Asymmetric Generalization Problem(tool)

Related

Paper

arXiv paper benchmarks LLM judges for citation quality in deep-research systems

A new arXiv paper studies the calibration of LLM judges used as reward models in reinforcement learning for citation quality in deep-research systems. The work evaluates how capable and biased an LLM judge must be to reliably score rubric criteria like source relevance and factual support for attribution-citation pairs. This matters for practitioners building RL-based systems that depend on automated citation verification.

0 engagement·1 source·arxiv
Thu, Jul 9, 2026, 05:01 PM
arXiv
Paper

Self-Guided Test-Time Training Improves Long-Context LLM Accuracy

A new arxiv paper proposes Self-Guided Test-Time Training (SG-TTT) to improve long-context utilization in LLMs without requiring labeled data. The method uses the model's own predictions to generate pseudo-labels for fine-tuning on the test context, addressing accuracy degradation in long inputs. This approach is more efficient than full TTT and shows promise for practical deployment.

0 engagement·1 source·arxiv
Fri, Jul 10, 2026, 01:45 PM
arXiv
Paper

NL-PAC paper introduces framework for handling specification ambiguity in LLM supervision

A new arXiv paper proposes NL-PAC, a framework that addresses specification ambiguity when LLMs provide labels or evaluations from natural language instructions. The framework uses a fixed model's thresholded decoding law to define admissible labels and candidate targets, showing that additional labels reduce sampling error but cannot resolve identification problems from ambiguous specifications.

0 engagement·1 source·arxiv
Thu, Jul 9, 2026, 09:53 PM
arXiv
Paper

Researchers identify modality interference as root cause of full-duplex SLM degradation

A new paper on arXiv (July 7, 2026) presents a fine-grained analysis of optimization dynamics in full-duplex Spoken Language Models (SLMs), identifying severe modality interference as the root cause of knowledge degradation and compromised semantic integrity. The work aims to enable more natural and intelligent full-duplex SLMs.

0 engagement·1 source·arxiv
Tue, Jul 7, 2026, 05:43 PM
arXiv
Paper

Study questions robustness of emergent misalignment in language models

A new paper systematically studies repeated alignment and misalignment cycles in language models, reproducing Emergent Misalignment (EM) but finding that both misalignment and realignment are highly sensitive to superficial dataset characteristics. The results suggest that EM may not be a robust phenomenon.

0 engagement·1 source·arxiv
Fri, Jul 10, 2026, 02:50 AM
arXiv