Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

GPT-2 embeddings

GPT-2 embeddings are high-dimensional vector representations of tokens or text sequences generated by the Generative Pre-trained Transformer 2 language model to capture semantic and syntactic information. Unlike traditional static word embeddings that assign a single fixed numerical vector to a word regardless of how it is used, GPT-2 embeddings are contextualized representations that dynamically adjust based on surrounding textual context. Within the model architecture, an initial combination of static token and learned positional embeddings is transformed across successive causal transformer layers into increasingly context-specific hidden states. These dense vector representations capture nuanced grammatical and semantic relationships across preceding context, allowing computational models to measure semantic similarity, interpret linguistic patterns, and perform downstream natural language processing tasks.

1 item

How Contextual are Contextualized Word Representations? Comparing the Geometry of BERT, ELMo, and GPT-2 Embeddings

How Contextual are Contextualized Word Representations? Comparing the Geometry of BERT, ELMo, and GPT-2 Embeddings

Kawin Ethayarajh

OrganizationsStanford University

Why you should read this

Reveals the geometric behavior of BERT, ELMo, and GPT-2 embeddings across layers, proving that upper layers generate more context-specific representations while static word vectors account for less than five percent of their variance.

Replacing static word embeddings with contextualized word representations has yielded significant improvements on many NLP tasks. However, just how contextual are the contextualized representations produced by models such as ELMo and BERT? Are there infinitely many context-specific representations for each word, or are words essentially assigned one of a finite number of word-sense representations? For one, we find that the contextualized representations of all words are not isotropic in any layer of the contextualizing model. While representations of the same word in different contexts still have a greater cosine similarity than those of two different words, this self-similarity is much lower in upper layers. This suggests that upper layers of contextualizing models produce more context-specific representations, much like how upper layers of LSTMs produce more task-specific representations. In all layers of ELMo, BERT, and GPT-2, on average, less than 5% of the variance in a word's contextualized representations can be explained by a static embedding for that word, providing some justification for the success of contextualized representations.

Added

2026-09-25