稀疏讀出稜鏡：以特徵而非令牌解釋Logit-Lens分數

摘要

語言模型對下一個詞元的預測會逐層發展成形，透鏡方法透過將中間隱藏狀態解碼為詞元來追蹤此過程。然而，透鏡的讀取結果同時反映了隱藏狀態以及解碼所用的讀出（反嵌入）矩陣。許多透鏡是在語料庫上擬合的，而我們證明，僅在擬合語料庫上有所不同的兩個透鏡，對相同的隱藏狀態可能報告不同的詞元。我們將此種依賴性稱為語料庫條件性。為了獨立於擬合語料庫檢驗讀出結構，我們提出稀疏讀出稜鏡（Sparse Readout Prism, SRP），該方法僅利用讀出矩陣的權重對其進行分解，並將任何詞元 logit 或 logit 差異表示為稀疏讀出特徵貢獻的總和。這使讀出特徵成為透鏡讀取分析的新分析單元，揭露了詞元身份可能掩蓋的結構，並促成跨詞元、跨上下文、跨層與跨透鏡的比較。以 SRP 的稀疏近似取代原始讀出矩陣，對受測 logit 差異的重建比例，比基於讀出矩陣各行間幾何關係所建構的六種基線中最強者高出 8.9–17.3 個百分點。消融特徵會使 logit 差異依其 SRP 貢獻的比例發生偏移。儘管詞元讀取結果隨擬合語料庫而變化，主導的讀出特徵仍保持穩定。由於 SRP 在建構過程中不使用任何語料庫，它為透鏡分析提供了獨立於擬合語料庫的對照。

English

A language model's prediction of its next token develops across layers, and lens methods track this process by decoding intermediate hidden states into tokens. But a lens reading reflects both the hidden state and the readout (the unembedding matrix) used to decode it. Many lenses are fit on a corpus, and we show that two lenses differing only in their fitting corpus can report different tokens for the same hidden states. We call this dependence corpus conditionality. To examine readout structure independently of the fitting corpus, we introduce Sparse Readout Prism (SRP), which decomposes the readout using only its weights and expresses any token logit or logit difference as a sum of contributions from sparse readout features. This reveals readout features as a new unit of analysis for lens readings, exposing structure that token identities can obscure and enabling comparisons across tokens, contexts, layers, and lenses. Replacing the original readout with SRP's sparse approximation reconstructs 8.9-17.3 percentage points more of the tested logit differences than the strongest of six baselines built on geometric relations among readout rows. Ablating features shifts logit differences in proportion to their SRP contributions. Although token readings vary with the fitting corpus, the dominant readout feature remains stable. Because SRP uses no corpus in its construction, it provides a control independent of the fitting corpus for lens analyses.