DataRx: Missingness-Aware Sampling for Safer Large Language Model Task-Specific Fine-Tuning
Junbo Zhang, Qianli Zhou, Xinyang Deng, Wen Jiang
cs.CL
Submitted: 2026-08-05
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- The Llama 3 Herd of Models
- GPT-4 Technical Report
- DataShield: Safety-degrading Data Filtering for LLM Benign Instruction Fine-Tuning
- Token-Level Adversarial Prompt Detection Based on Perplexity Measures and Contextual Information
- GR-SAP: Generative Replay for Safety Alignment Preservation during Fine-Tuning
- Qwen2.5 Technical Report
- Mistral 7B
- Training Verifiers to Solve Math Word Problems
- Measuring Mathematical Problem Solving With the MATH Dataset
- Multilingual Contextualization of Large Language Models for Document-Level Machine Translation
- Evaluating Large Language Models Trained on Code
- HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal
- Long Is More for Alignment: A Simple but Tough-to-Beat Baseline for Instruction Fine-Tuning
- Qwen3Guard Technical Report
- Understanding and Mitigating Over-refusal for Large Language Models via Safety Representation
- Unlearning Isn't Deletion: Investigating Reversibility of Machine Unlearning in LLMs
- LIMO: Less is More for Reasoning
- Hint Tuning: Less Data Makes Better Reasoners
- Why LLM Safety Guardrails Collapse After Fine-tuning: A Similarity Analysis Between Alignment and Fine-tuning Datasets
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering