The paper investigates how semantic meaning is arranged locally within large language models. By using two controlled probes—one cross‑lingual and one refinement probe—on four decoder‑only LLMs, it finds that after removing dominant variance directions, sentence pairs rank by semantic relatedness, semantic factors cluster around an invariant core with structured variation, refinement tightens neighborhoods, and sentence‑level versus token‑local readouts reveal complementary geometric patterns. These results suggest that LLM semantics form local anisotropic neighborhoods that contract when meaning is refined.
arXiv:2605.01609v2 Announce Type: replace-cross Abstract: We find that transformer concept representations systematically anti-concentrate in the spectral tail of the unembedding covariance, encoding...
arXiv:2408.11827v2 Announce Type: replace Abstract: Understanding how language models compose meaning from linguistic input remains a central problem in interpretability research. Mechanistic studies...
arXiv:2610.14476v1 Announce Type: cross Abstract: Comparing large language models can reveal shared representational structure and systematic differences between them. However, aggregate similarity m...
arXiv:2603. 04419v2 Announce Type: replace-cross Abstract: We characterize the phenomenon of context-dependent affordance computation in vision-language models (VLMs).
The paper extends mechanistic interpretability of large language models by modeling concepts as low‑dimensional non‑linear manifolds rather than linear subspaces. It introduces a concept‑based alignment (CBA) score to compare these manifolds across layers and models, revealing block structures in intermediate layers, a shift from syntax‑dominated to mixed syntactic‑semantic concepts, and training‑dependent multilingual sharing. The study also shows that alignment patterns differ across model families and training stages, with adjacent stages aligning more closely than distant ones.
The study investigates why language models exhibit systematic performance gaps across English dialects, a phenomenon termed the "dialect tax." Using parallel dialect corpora that preserve meaning while altering surface form, the authors confirm that models treat Standard American English and dialectal texts as semantically equivalent, yet find representational disparities that persist through tokenization, pre‑training, post‑training, and inference. Even a character‑level tokenizer does not eliminate input/output asymmetries or accuracy gaps, and dialect pairs produce more divergent gradient updates than unrelated Standard texts, indicating that dialectal content is harder for models to learn.
The paper compares latent representations in Selective State Space Models (SSMs) like Mamba and Transformers such as Pythia using Sparse Autoencoders. Across a 10‑million token corpus, 99.98% of Mamba features align closely with Pythia’s, supporting the Universality Hypothesis that core semantic representations are similar across architectures. A tiny 0.02% of features diverge, with Mamba’s recurrent bottleneck causing it to compress syntactic anomalies into polysemantic neurons, whereas Pythia’s attention can isolate distinct formatting edge‑cases.
arXiv:2609.34187v2 Announce Type: replace-cross Abstract: The strong version of the stochastic parrot argument claims that, although large language models (LLMs) may exceed rote regurgitation, they c...
The paper introduces LLM-Microscope, a toolkit for measuring how Large Language Models encode contextual information at the token level. It shows that seemingly minor tokens—such as determiners, stopwords, and punctuation—carry surprisingly high contextual weight, and removing them degrades performance on benchmarks like MMLU and BABILong-4k. The study also finds a strong link between contextualization and linearity, indicating that the transformation between layers can be approximated by a single linear mapping when tokens are well contextualized.
arXiv:2610.08303v1 Announce Type: new Abstract: Current evaluation of multilingual Large Language Models (LLMs) rests on an implicit Translation-Isomorphism Assumption (TIA): that semantic structures...
The article presents a technical manual for an open toolkit designed to measure how transformer language models individuate word meanings across different contexts. It introduces the concept of a "bridge form"—a single word that appears unchanged in multiple domains but with distinct senses—and outlines a full pipeline from specifying these forms to extracting layer-wise representations, computing silhouette-based separation metrics, and visualizing results. The manual details each design choice and its intended methodological safeguards, emphasizing that it serves as a methodological reference rather than reporting empirical findings.
arXiv:2610.03551v1 Announce Type: new Abstract: Large language models (LLMs) have reached expert-level performance on competition mathematics largely through the volume of search placed around them:...