Unverified50% confidenceFactExact time
Activation steering is a technique that intervenes in the model's internal hidden-layer representations at inference time, with representative methods including contrastive activation addition (CAA) and inference-time intervention (ITI).
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIntrospective instructions are essentially a metacognitive prompting technique, an extension of chain-of-thought in Prompt Engineering, whose core principle is to force the model to add a verification step before outputting a conclusion.69% similarUnverifiedThe idea of controlled interventions originates from do-calculus in the field of causal inference.69% similarUnverifiedInference deployment refers to actually running a trained or fine-tuned model to provide services externally, involving engineering problems such as model quantization, VRAM optimization, concurrency scheduling, batching, and the latency-throughput trade-off.65% similarUnverifiedActivation steering or representation engineering applies intervention vectors directly to the model's internal activation layers, altering the generation direction without modifying the model's weights.65% similarUnverifiedInference-Time Alignment is an umbrella term for a class of techniques that do not require retraining model parameters and only apply interventions during the generation stage, including prompt engineering, activation steering, decoding constraints, and other means.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1038556API
curl https://kongchang.com/api/v1/knowledge/claims/1038556MCP
get_claim(id=1038556)