flâneur

[2509.14223] Fresh in memory: Training-order recency is linearly encoded in language model activations

arxiv.org · 698 words · saved by 1 readers

Abstract:We show that language models' activations linearly encode when information was learned during training. Our setup involves creating a model with a known training order by sequentially fine-tuning Llama-3.2-1B on six disjoint but otherwise similar datasets about named entities. We find that the average activations of test samples corresponding to the six training datasets encode the training order: when projected into a 2D subspace, these centroids are arranged exactly in the order of training and lie on a straight line. Further, we show that linear probes can accurately (~90%) distinguish "early" vs. "late" entities, generalizing to entities unseen during the probes' own training. The model can also be fine-tuned to explicitly report an unseen entity's training stage (~80% accuracy). Interestingly, the training-order encoding does not seem attributable to simple differences in activation magnitudes, losses, or model confidence. Our paper demonstrates that models are capable of differentiating information by its acquisition time, and carries significant implications for how they might manage conflicting data and respond to knowledge modifications.

[2509.14223] Fresh in memory: Training-order recency is linearly encoded in language model activations Skip to main content arXiv is now an independent nonprofit! Learn more × Search arXiv Press Enter to search · Advanced search --> Computer Science > Machine Learning arXiv:2509.14223 (cs) [Submitted on 17 Sep 2025 ( v1 ), last revised 22 Sep 2025 (this version, v2)] Title: Fresh in memory: Training-order recency is linearly encoded in language model activations Authors: Dmitrii Krasheninnikov , Richard E. Turner , David Krueger View a PDF of the paper titled Fresh in memory: Trai

related reading