语言模型能线性编码训练顺序，揭示信息学习时间。
Fresh in memory: Training-order recency is linearly encoded in language model activations
- 通过顺序微调构建已知训练顺序的模型
- 激活中心点在二维空间中呈直线排列，准确率约90%
- 可识别未见实体的训练阶段，适合研究知识演化
我们发现语言模型的激活值线性编码了信息在训练中的学习时间。实验通过将Llama-3.2-1B模型依次在六个互不重叠但相似的命名实体数据集上微调，建立已知训练顺序。测试样本的平均激活值对应六个数据集的中心点，在二维子空间中严格按训练顺序排列成直线。线性探测器能以约90%准确率区分‘早期’与‘晚期’实体，且泛化至训练中未见的实体。模型还可微调以直接预测未见实体的训练阶段，准确率达80%。该编码现象并非由激活幅度、损失或置信度等简单差异导致。论文表明模型能区分信息获取时间，对处理冲突数据和知识更新具有重要启示。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。