TARDIS: Mitigating Temporal Misalignment via Representation Steering
On the map: Mechanistic interpretability → Activation steering methods
Abstract (arXiv)
Language models often struggle with temporal misalignment, performance degradation caused by shifts in the temporal distribution of data. Continuously updating models to avoid degradation is expensive. Can models be adapted without updating model weights? We present TARDIS, an unsupervised representation editing method that addresses this challenge. TARDIS extracts steering vectors from unlabeled data and adjusts the model's representations to better align with the target time period's distribution. Our experiments reveal that TARDIS enhances downstream task performance without the need for fine-tuning, can mitigate temporal misalignment even when exact target time period data is unavailable, and remains efficient even when the temporal information of the target data points is unknown at inference time.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability. Builds on 12 of its 24 references that are in the map. First posted on arXiv 24 Mar 2025.
Builds on
12 of 24 references.
References · 12
- 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
- 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,307
- 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,336
- 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,302
- 2023In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space Steeringcited by 245
- 2024A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicitycited by 185
- 2023Linear Representations of Sentiment in Large Language Modelscited by 163
- 2021Time Waits for No One! Analysis and Challenges of Temporal Misalignmentcited by 91
- 2021Temporal Adaptation of BERT and Performance on Downstream Document Classification: Insights from Social Mediacited by 51
- 2023Time is Encoded in the Weights of Finetuned Language Modelscited by 33
- 2024Discovering Bias in Latent Space: An Unsupervised Debiasing Approachcited by 15
- 2023Time-Aware Representation Learning for Time-Sensitive Question Answeringcited by 7
Built on by
Citing papers per year: 2026: 1.
1 paper.
Similar, not cited
Close in wording, no citation either way
- 2024Aligning Large Language Models with Representation Editing: A Control Perspective
- 2026Steer2Edit: From Activation Steering to Component-Level Editing
- 2026Activation Steering Induces Emergent Misalignment: A More Comprehensive Evaluation
- 2025Temporal Alignment of LLMs through Cycle Encoding for Long-Range Time Representations
- 2025Self-Improving Model Steering
- 2024Nudging: Inference-time Alignment of LLMs via Guided Decoding
- 2025Aligning Language Models with Real-time Knowledge Editing
- 2026Where to Steer: Input-Dependent Layer Selection for Steering Improves LLM Alignment