Activation Scaling for Steering and Interpreting Language Models

• URL: http://arxiv.org/abs/2410.04962v1
• Date: Mon, 7 Oct 2024 12:01:32 GMT
• Title: Activation Scaling for Steering and Interpreting Language Models
• Authors: Niklas Stoehr, Kevin Du, Vésteinn Snæbjarnarson, Robert West, Ryan Cotterell, Aaron Schein,
• Abstract summary: We argue that successfully intervening on a model is a prerequisite for interpreting its internal workings. We establish a three-term objective: a successful intervention should flip the correct with the wrong token and vice versa. Using gradient-based optimization, this objective lets us learn (and later evaluate) a specific kind of efficient and interpretable intervention.
• Score: 55.59689963561315
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Given the prompt "Rome is in", can we steer a language model to flip its prediction of an incorrect token "France" to a correct token "Italy" by only multiplying a few relevant activation vectors with scalars? We argue that successfully intervening on a model is a prerequisite for interpreting its internal workings. Concretely, we establish a three-term objective: a successful intervention should flip the correct with the wrong token and vice versa (effectiveness), and leave other tokens unaffected (faithfulness), all while being sparse (minimality). Using gradient-based optimization, this objective lets us learn (and later evaluate) a specific kind of efficient and interpretable intervention: activation scaling only modifies the signed magnitude of activation vectors to strengthen, weaken, or reverse the steering directions already encoded in the model. On synthetic tasks, this intervention performs comparably with steering vectors in terms of effectiveness and faithfulness, but is much more minimal allowing us to pinpoint interpretable model components. We evaluate activation scaling from different angles, compare performance on different datasets, and make activation scalars a learnable function of the activation vectors themselves to generalize to varying-length prompts.

Related papers

• Improving Instruction-Following in Language Models through Activation Steering [58.876600545898675]
  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.
  arXiv  Detail & Related papers  (2024-10-15T08:38:20Z)
• An Energy-based Model for Word-level AutoCompletion in Computer-aided Translation [97.3797716862478]
  Word-level AutoCompletion (WLAC) is a rewarding yet challenging task in Computer-aided Translation. Existing work addresses this task through a classification model based on a neural network that maps the hidden vector of the input context into its corresponding label. This work proposes an energy-based model for WLAC, which enables the context hidden vector to capture crucial information from the source sentence.
  arXiv  Detail & Related papers  (2024-07-29T15:07:19Z)
• A Pattern Language for Machine Learning Tasks [0.0]
  We view objective functions as constraints on the behaviour of learners. We develop a formal graphical language that allows us to separate the core tasks of a behaviour from its implementation details. As proof-of-concept, we design a novel task that enables converting classifiers into generative models we call "manipulators"
  arXiv  Detail & Related papers  (2024-07-02T16:50:27Z)
• 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)
• Steering Language Models With Activation Engineering [40.04138190785384]
  We introduce activation engineering: the inference-time modification of activations in order to control (or steer) model outputs. We achieve SOTA on negative-to-positive sentiment shift and detoxification using models including LLaMA-3 and OPT. ActAdd yields inference-time control over high-level output properties (like topic and sentiment) while preserving performance on off-target tasks.
  arXiv  Detail & Related papers  (2023-08-20T12:21:05Z)
• Interpretability at Scale: Identifying Causal Mechanisms in Alpaca [62.65877150123775]
  We use Boundless DAS to efficiently search for interpretable causal structure in large language models while they follow instructions. Our findings mark a first step toward faithfully understanding the inner-workings of our ever-growing and most widely deployed language models.
  arXiv  Detail & Related papers  (2023-05-15T17:15:40Z)
• VECO 2.0: Cross-lingual Language Model Pre-training with Multi-granularity Contrastive Learning [56.47303426167584]
  We propose a cross-lingual pre-trained model VECO2.0 based on contrastive learning with multi-granularity alignments. Specifically, the sequence-to-sequence alignment is induced to maximize the similarity of the parallel pairs and minimize the non-parallel pairs. token-to-token alignment is integrated to bridge the gap between synonymous tokens excavated via the thesaurus dictionary from the other unpaired tokens in a bilingual instance.
  arXiv  Detail & Related papers  (2023-04-17T12:23:41Z)
• Sentence Representation Learning with Generative Objective rather than Contrastive Objective [86.01683892956144]
  We propose a novel generative self-supervised learning objective based on phrase reconstruction. Our generative learning achieves powerful enough performance improvement and outperforms the current state-of-the-art contrastive methods.
  arXiv  Detail & Related papers  (2022-10-16T07:47:46Z)
• Extracting Latent Steering Vectors from Pretrained Language Models [14.77762401765532]
  We show that latent vectors can be extracted directly from language model decoders without fine-tuning. Experiments show that there exist steering vectors, which, when added to the hidden states of the language model, generate a target sentence nearly perfectly. We find that distances between steering vectors reflect sentence similarity when evaluated on a textual similarity benchmark.
  arXiv  Detail & Related papers  (2022-05-10T19:04:37Z)
• ASFormer: Transformer for Action Segmentation [9.509416095106493]
  We present an efficient Transformer-based model for action segmentation task, named ASFormer. It constrains the hypothesis space within a reliable scope, and is beneficial for the action segmentation task to learn a proper target function with small training sets. We apply a pre-defined hierarchical representation pattern that efficiently handles long input sequences.
  arXiv  Detail & Related papers  (2021-10-16T13:07:20Z)

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.