ASA: Backbone-Training-Free Representation Engineering for Tool-Calling Agents
Adapting LLM agents to domain-specific tool calling remains notably brittle under evolving interfaces. Prompt and schema engineering is easy to deploy but often fragile under distribution shift and strict parsers, while continual parameter-efficient fine-tuning improves reliability at the cost of training, maintenance, and potential forgetting. We identify a critical Lazy Agent failure mode where tool necessity is nearly perfectly decodable from mid-layer activations, yet the model remains conservative in entering tool mode, revealing a representation-behavior gap. We propose Activation Steering Adapter (ASA), a training-free, inference-time controller that performs a single-shot mid-layer intervention and targets tool domains via a router-conditioned mixture of steering vectors with a probe-guided signed gate to amplify true intent while suppressing spurious triggers. On MTU-Bench with Qwen2.5-1.5B, ASA improves strict tool-use F1 from 0.18 to 0.50 while reducing the false positive rate from 0.15 to 0.05, using only about 20KB of portable assets and no weight updates.
Code (0)
등록된 구현이 없습니다.
Tasks
parameter-efficient fine-tuningSimilar Papers 제목 키워드 기반
SkillGPT: a RESTful API service for skill extraction and standardization using a Large Language Model
We present SkillGPT, a tool for skill extraction and standardization (SES) from free-style job descriptions and user profiles with an open-source Large Language Model (LLM) as backbone. Most previous methods for similar …
Feature EngineeringLanguage ModelingLanguage ModellingLarge Language ModelUnlocking Multilingual Reasoning Capability of LLMs and LVLMs through Representation Engineering
Large Language Models (LLMs) and Large Vision-Language Models (LVLMs) demonstrate strong reasoning capabilities, yet their performance in English significantly outperforms that in low-resource languages, raising fairness…
Dynamo: Dynamic Skill-Tool Evolution for Vision-Language Agents
Improving vision-language models (VLMs) on visual reasoning typically requires retraining or hand-designed prompts and tools. We present Dynamo, a training-free framework that adapts a frozen VLM without any weight updat…
Visual ReasoningExperimental Demonstration of an Optical Neural PDE Solver via On-Chip PINN Training
Partial differential equation (PDE) is an important math tool in science and engineering. This paper experimentally demonstrates an optical neural PDE solver by leveraging the back-propagation-free on-photonic-chip train…
MathMachine learning and control engineering: The model-free case
This paper states that Model-Free Control (MFC), which must not be confused with Model-Free Reinforcement Learning, is a new tool for Machine Learning (ML). MFC is easy to implement and should be substituted in control e…
BIG-bench Machine Learningreinforcement-learningReinforcement LearningReinforcement Learning (RL)