MPSelectTune: Prompt-type Selection for Fine-tuning improves Concept Unlearning in LLMs

• URL: http://arxiv.org/abs/2607.03932v1
• Date: Sat, 04 Jul 2026 15:52:16 GMT
• Title: MPSelectTune: Prompt-type Selection for Fine-tuning improves Concept Unlearning in LLMs
• Abstract summary: Biased or harmful concepts, e.g. gender or bio-weapons, can lead to unsafe or unethical responses for many such prompts.<n>We show that fine-tuning using the worst prompt type'' for concept prediction (with the highest concept accuracy) improves the average unlearning performance.<n>Our proposed method, MPSelectTune, is a two-stage approach that minimizes the concept accuracy of the highest accuracy-prompt type, after fine-tuning using a novel multi-task loss.
• Score: 5.054661352089593
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: LLMs can be conveniently adapted to a diverse set of tasks, e.g, prediction, question-answering tasks, etc, using appropriate prompts with few-shot examples. Biased or harmful concepts, e.g. gender or bio-weapons, present in pre-trained LLMs can lead to unsafe or unethical responses for many such prompts. Removing such undesirable concepts robustly across different prompt types remains a challenging problem, since existing unlearning methods typically ignore the impact of prompt variation. In this paper, we explore a novel adversarial approach to use a joint prompt for the main task and concept task prediction. We show that fine-tuning using the ``worst prompt type'' for concept prediction (with the highest concept accuracy) improves the average unlearning performance over a fine-tuning method that uses a combination of all prompt types. Our proposed method, MPSelectTune, is a two-stage approach that minimizes the concept accuracy of the highest accuracy-prompt type, after fine-tuning using a novel multi-task loss using multiple prompt types. Experimental results on four benchmarks show $2 - 15\%$ main task accuracy improvements over recent baselines and while reducing the worst-case concept accuracy by up to $17\%$ compared to recent baselines.

Related papers

• $p1$: Better Prompt Optimization with Fewer Prompts [49.20082664169319]
  We show that prompt optimization succeeds when variance among system prompts is large, but fails when variance among responses dominates the variance of the system prompts.<n>We propose $p1$, a simple user prompt filtering method that selects a small subset of user prompts with high variance across candidate system prompts.
  arXiv  Detail & Related papers  (2026-04-09T22:31:15Z)
• CAPrompt: Cyclic Prompt Aggregation for Pre-Trained Model Based Class Incremental Learning [12.249938312431993]
  We propose a novel Cyclic Prompt Aggregation (CAPrompt) method to eliminate the dependency on task ID prediction.<n>Under concave conditions, the aggregated prompt achieves lower error compared to selecting a single task-specific prompt.<n>Our proposed CAPrompt outperforms state-of-the-art methods by 2%-3%.
  arXiv  Detail & Related papers  (2024-12-12T04:34:28Z)
• Vector Quantization Prompting for Continual Learning [23.26682439914273]
  Continual learning requires to overcome catastrophic forgetting when training a single model on a sequence of tasks.<n>Recent top-performing approaches are prompt-based methods that utilize a set of learnable parameters to encode task knowledge.<n>We propose VQ-Prompt, a prompt-based continual learning method that incorporates Vector Quantization into end-to-end training of a set of discrete prompts.
  arXiv  Detail & Related papers  (2024-10-27T13:43:53Z)
• Efficient Prompt Tuning by Multi-Space Projection and Prompt Fusion [9.668242919588199]
  Prompt tuning is a promising method to fine-tune a pre-trained language model without retraining its large-scale parameters.<n>Existing methods are hard to balance accuracy and efficiency.<n>A longer (shorter) soft prompt generally leads to a better(worse) accuracy but at the cost of more (less) training time.<n>We propose an Efficient Prompt Tuning method (EPT) by multi-space projection and prompt fusion.
  arXiv  Detail & Related papers  (2024-05-19T06:43:12Z)
• OVOR: OnePrompt with Virtual Outlier Regularization for Rehearsal-Free Class-Incremental Learning [9.164352588061075]
  We propose a regularization method based on virtual outliers to tighten decision boundaries of the classifier.<n>A simplified prompt-based method can achieve results comparable to previous state-of-the-art (SOTA) methods equipped with a prompt pool.
  arXiv  Detail & Related papers  (2024-02-06T16:31:11Z)
• Self-regulating Prompts: Foundational Model Adaptation without Forgetting [112.66832145320434]
  We introduce a self-regularization framework for prompting called PromptSRC. PromptSRC guides the prompts to optimize for both task-specific and task-agnostic general representations.
  arXiv  Detail & Related papers  (2023-07-13T17:59:35Z)
• Multitask Prompt Tuning Enables Parameter-Efficient Transfer Learning [43.639430661322585]
  We propose multitask prompt tuning (MPT) MPT learns a single transferable prompt by distilling knowledge from multiple task-specific source prompts. We then learn multiplicative low rank updates to this shared prompt to efficiently adapt it to each downstream target task.
  arXiv  Detail & Related papers  (2023-03-06T03:25:59Z)
• TEMPERA: Test-Time Prompting via Reinforcement Learning [57.48657629588436]
  We propose Test-time Prompt Editing using Reinforcement learning (TEMPERA) In contrast to prior prompt generation methods, TEMPERA can efficiently leverage prior knowledge. Our method achieves 5.33x on average improvement in sample efficiency when compared to the traditional fine-tuning methods.
  arXiv  Detail & Related papers  (2022-11-21T22:38:20Z)
• LASP: Text-to-Text Optimization for Language-Aware Soft Prompting of Vision & Language Models [67.19124099815645]
  We propose a novel Language-Aware Soft Prompting (LASP) learning method to alleviate base class overfitting. LASP is inherently amenable to including, during training, virtual classes, i.e. class names for which no visual samples are available. LASP matches and surpasses, for the first time, the accuracy on novel classes obtained by hand-crafted prompts and CLIP for 8 out of 11 test datasets.
  arXiv  Detail & Related papers  (2022-10-03T17:56:35Z)
• Interactive and Visual Prompt Engineering for Ad-hoc Task Adaptation with Large Language Models [116.25562358482962]
  State-of-the-art neural language models can be used to solve ad-hoc language tasks without the need for supervised training. PromptIDE allows users to experiment with prompt variations, visualize prompt performance, and iteratively optimize prompts.
  arXiv  Detail & Related papers  (2022-08-16T17:17:53Z)
• RLPrompt: Optimizing Discrete Text Prompts With Reinforcement Learning [84.75064077323098]
  This paper proposes RLPrompt, an efficient discrete prompt optimization approach with reinforcement learning (RL) RLPrompt is flexibly applicable to different types of LMs, such as masked gibberish (e.g., grammaBERT) and left-to-right models (e.g., GPTs) Experiments on few-shot classification and unsupervised text style transfer show superior performance over a wide range of existing finetuning or prompting methods.
  arXiv  Detail & Related papers  (2022-05-25T07:50:31Z)
• Calibrate Before Use: Improving Few-Shot Performance of Language Models [68.17016463756474]
  GPT-3 can perform numerous tasks when provided a natural language prompt that contains a few training examples. We show that this type of few-shot learning can be unstable. The choice of prompt format, training examples, and even the order of the training examples can cause accuracy to vary from near chance to near state-of-the-art.
  arXiv  Detail & Related papers  (2021-02-19T00:23:59Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.