Formula

Input Embedding Formula in BERT-like Models

In BERT models, the input is a sequence of embeddings, where each individual embedding, denoted as e\mathbf{e}, is the sum of the token embedding (x\mathbf{x}), the positional embedding (epos\mathbf{e}_{\mathrm{pos}}), and the segment embedding (eseg\mathbf{e}_{\mathrm{seg}}). The mathematical formula for this composition is: e=x+epos+eseg\mathbf{e} = \mathbf{x} + \mathbf{e}_{\mathrm{pos}} + \mathbf{e}_{\mathrm{seg}}.

0

1

Updated 2026-04-17

Contributors are:

Who are from:

Tags

Ch.1 Pre-training - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences

Related