Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

COLD-Steer: Steering Large Language Models via In-Context One-step Learning Dynamics

About

Activation steering methods enable inference-time control of large language model (LLM) behavior without retraining, but current approaches face a fundamental trade-off: sample-efficient methods suboptimally capture steering signals from labeled examples, while methods that better extract these signals require hundreds to thousands of examples. We introduce COLD-Steer, a training-free framework that steers LLM activations by approximating the representational changes that would result from gradient descent on in-context examples. Our key insight is that the effect of fine-tuning on a small set of examples can be efficiently approximated at inference time without actual parameter updates. We formalize this through two complementary approaches: (i) a unit kernel approximation method that updates the activations directly using gradients with respect to them, normalized across examples, and (ii) a finite-difference approximation requiring only two forward passes regardless of example count. Experiments across a variety of steering tasks and benchmarks demonstrate that COLD-Steer achieves upto 95% steering effectiveness while using 50 times fewer samples compared to the best baseline. COLD-Steer facilitates accommodating diverse perspectives without extensive demonstration data, which we validate through our experiments on pluralistic alignment tasks. Our framework opens new possibilities for adaptive, context-aware model control that can flexibly address varying loss-driven human preferences through principled approximation of learning dynamics rather than specialized training procedures.

Kartik Sharma, Rakshit S. Trivedi• 2026

Related benchmarks

TaskDatasetResultRank
Behavior selectionCAA (50 random samples)
Accuracy (coordinate-ais, pair)98
22
Behavior selectionBiPO (test)
Hallucination Pair Accuracy64
14
Hallucination SteeringCAA
Runtime31.14
13
Open-ended behavior generationCAA
CoAIS Score4.36
10
HallucinationCAA
Accuracy (pair)88
8
Behavior selectionCAA behaviors
COAIS Score4.64
5
Behavior GenerationBiPO
Hallucination Score1.62
5
Behavior GenerationCAA behaviors Qwen-2.5-7B-Instruct (test)
CoAIS26
5
Distributional pluralistic alignmentOpinionsQA
Democrat KL Divergence1
4
Showing 9 of 9 rows

Other info

Follow for update