Improving Instruction-Following in Language Models through Activation Steering

• URL: http://arxiv.org/abs/2410.12877v1
• Date: Tue, 15 Oct 2024 08:38:20 GMT
• Title: Improving Instruction-Following in Language Models through Activation Steering
• Authors: Alessandro Stolfo, Vidhisha Balachandran, Safoora Yousefi, Eric Horvitz, Besmira Nushi,
• Abstract summary: We derive instruction-specific vector representations from language models and use them to steer models accordingly. We demonstrate how this method can enhance model adherence to constraints such as output format, length, and word inclusion. Our findings demonstrate that activation steering offers a practical and scalable approach for fine-grained control in language generation.
• Score: 58.876600545898675
• License:
• Abstract: The ability to follow instructions is crucial for numerous real-world applications of language models. In pursuit of deeper insights and more powerful capabilities, we derive instruction-specific vector representations from language models and use them to steer models accordingly. These vectors are computed as the difference in activations between inputs with and without instructions, enabling a modular approach to activation steering. We demonstrate how this method can enhance model adherence to constraints such as output format, length, and word inclusion, providing inference-time control over instruction following. Our experiments across four models demonstrate how we can use the activation vectors to guide models to follow constraints even without explicit instructions and to enhance performance when instructions are present. Additionally, we explore the compositionality of activation steering, successfully applying multiple instructions simultaneously. Finally, we demonstrate that steering vectors computed on instruction-tuned models can transfer to improve base models. Our findings demonstrate that activation steering offers a practical and scalable approach for fine-grained control in language generation.

Related papers

• Steering Large Language Models using Conceptors: Improving Addition-Based Activation Engineering [0.0]
  This paper explores activation engineering, where outputs of pre-trained LLMs are controlled by manipulating their activations at inference time. We introduce conceptors - mathematical constructs that represent sets of activation vectors as ellipsoidal regions. Our experiments demonstrate that conceptors outperform traditional methods across multiple in-context learning steering tasks.
  arXiv  Detail & Related papers  (2024-10-09T10:09:37Z)
• Controllable Navigation Instruction Generation with Chain of Thought Prompting [74.34604350917273]
  We propose C-Instructor, which utilizes the chain-of-thought-style prompt for style-controllable and content-controllable instruction generation. C-Instructor renders generated instructions more accessible to follow and offers greater controllability over the manipulation of landmark objects.
  arXiv  Detail & Related papers  (2024-07-10T07:37:20Z)
• Fine-Tuning with Divergent Chains of Thought Boosts Reasoning Through Self-Correction in Language Models [63.36637269634553]
  We present a novel method of further improving performance by requiring models to compare multiple reasoning chains. We find that instruction tuning on DCoT datasets boosts the performance of even smaller, and therefore more accessible, language models.
  arXiv  Detail & Related papers  (2024-07-03T15:01:18Z)
• Transformer-based Causal Language Models Perform Clustering [20.430255724239448]
  We introduce a simplified instruction-following task and use synthetic datasets to analyze a Transformer-based causal language model. Our findings suggest that the model learns task-specific information by clustering data within its hidden space, with this clustering process evolving dynamically during learning.
  arXiv  Detail & Related papers  (2024-02-19T14:02:31Z)
• Improving Activation Steering in Language Models with Mean-Centring [10.101141087916133]
  We find that taking the average of activations associated with a target dataset, and subtracting the mean of all training activations, results in effective steering vectors. We also apply mean-centring to extract function vectors, more effectively triggering the execution of a range of natural language tasks by a significant margin.
  arXiv  Detail & Related papers  (2023-12-06T18:27:07Z)
• Instruction-following Evaluation through Verbalizer Manipulation [64.73188776428799]
  We propose a novel instruction-following evaluation protocol called verbalizer manipulation. It instructs the model to verbalize the task label with words aligning with model priors to different extents. We observe that the instruction-following abilities of models, across different families and scales, are significantly distinguished by their performance on less natural verbalizers.
  arXiv  Detail & Related papers  (2023-07-20T03:54:24Z)
• Few-shot Prompting Towards Controllable Response Generation [49.479958672988566]
  We first explored the combination of prompting and reinforcement learning (RL) to steer models' generation without accessing any of the models' parameters. We apply multi-task learning to make the model learn to generalize to new tasks better. Experiment results show that our proposed method can successfully control several state-of-the-art (SOTA) dialogue models without accessing their parameters.
  arXiv  Detail & Related papers  (2022-06-08T14:48:06Z)
• Skill Induction and Planning with Latent Language [94.55783888325165]
  We formulate a generative model of action sequences in which goals generate sequences of high-level subtask descriptions. We describe how to train this model using primarily unannotated demonstrations by parsing demonstrations into sequences of named high-level subtasks. In trained models, the space of natural language commands indexes a library of skills; agents can use these skills to plan by generating high-level instruction sequences tailored to novel goals.
  arXiv  Detail & Related papers  (2021-10-04T15:36:32Z)

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.