PapersSelected73°

Training-Free Task Vectors for LLM Control

Training-Free Task Vectors for LLM Behavioral Control

Why we picked it

TFTVs reduce computational costs for post-training model editing, making it more practical for developers working with large language models.

Researchers introduced Training-Free Task Vectors (TFTVs) to compute task-vector-like directions without requiring fine-tuning.

Body

The method maps activation steering vectors to rank-one weight-space edits using only forward-pass statistics.

TFTVs consistently amplify, suppress, and compose target behaviors while preserving general knowledge and problem-solving skills.

The approach achieves stronger trait control with better or competitive utility preservation compared to other editing and steering baselines.

Original · arXiv cs.AI

Training-Free Task Vectors for LLM Behavioral Control

Task vectors enable post-training model editing by identifying semantically meaningful directions in weight space, typically computed as the difference between a fine-tuned model and its pretrained initialization. However, this reliance on fine-tuning makes discovering such directions costly and limits the practicality of post-training model editing. To address this limitation, we introduce Training-Free Task Vectors (TFTVs), a novel method to compute task-vector-like directions without requiring fine-tuning. Our method maps activation steering vectors to rank-one weight-space edits using only forward-pass statistics, while satisfying arithmetic properties that directly support learning via addition, forgetting via subtraction, and the composition of multiple edits. Empirically, we evaluate TFTVs on large language model behavioral control tasks and show that they consistently amplify, suppress, and compose target behaviors while preserving general knowledge and problem-solving skills. We also validate our method against other editing and steering baselines, experimentally demonstrating that TFTVs achieve stronger trait control with better or competitive utility preservation. We hope our work opens new directions for the community in post-training model editing and broader training-free model control. Code is available on the project website: tftv-llm.github.io.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include TFTVs, LLM, model editing.

Source attribution: curated from arXiv cs.AI (@Gabriel J. Perin, Lucas Boscaini, André Araujo, Nina S. T. Hirata).