Semantic Structure in Large Language Model Embeddings
Psychological research consistently finds that human ratings of words across diverse semantic scales can be reduced to a low-dimensional form with relatively little information loss. We find that the semantic associations encoded in the embedding matrices of large language models (LLMs) exhibit a similar structure. We show that the projections of words on semantic directions defined by antonym pairs (e.g. kind - cruel) correlate highly with human ratings, and further find that these projections effectively reduce to a 3-dimensional subspace within LLM embeddings, closely resembling the patterns derived from human survey responses. Moreover, we find that shifting tokens along one semantic direction causes off-target effects on geometrically aligned features proportional to their cosine similarity. These findings suggest that semantic features are entangled within LLMs similarly to how they are interconnected in human language, and a great deal of semantic information, despite its apparent complexity, is surprisingly low-dimensional. Furthermore, accounting for this semantic structure may prove essential for avoiding unintended consequences when steering features.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
From the New World of Word Embeddings: A Comparative Study of Small-World Lexico-Semantic Networks in LLMs
Lexico-semantic networks represent words as nodes and their semantic relatedness as edges. While such networks are traditionally constructed using embeddings from encoder-based models or static vectors, embeddings from d…
DecoderWord EmbeddingsDiscovering Multi-Scale Semantic Structure in Text Corpora Using Density-Based Trees and LLM Embeddings
Recent advances in large language models enable documents to be represented as dense semantic embeddings, supporting similarity-based operations over large text collections. However, many web-scale systems still rely on …
Vocabulary embeddings organize linguistic structure early in language model training
Large language models (LLMs) work by manipulating the geometry of input embedding vectors over multiple layers. Here, we ask: how are the input vocabulary representations of language models structured, and how and when d…
Semantic Structure and Interpretability of Word Embeddings
Dense word embeddings, which encode semantic meanings of words to low dimensional vector spaces have become very popular in natural language processing (NLP) research due to their state-of-the-art performances in many NL…
Word EmbeddingsEDS-MEMBED: Multi-sense embeddings based on enhanced distributional semantic structures via a graph walk over word senses
Several language applications often require word semantics as a core part of their processing pipeline, either as precise meaning inference or semantic similarity. Multi-sense embeddings (M-SE) can be exploited for this …
Semantic SimilaritySemantic Textual SimilarityWord Sense DisambiguationWord Similarity