---
schema: 1
type: term
title: '**training corpus**'
description: >-
  **training corpus** — as used in this channel's material, the large collection
  of text that a language model is shaped by; the source body of writing from
  which the model's sense of language and its outputs derive.
canonical: >-
  https://liquidgold.cc/channel/everyinc/explore/term/training-corpus-9f27feadcd19
markdown: >-
  https://liquidgold.cc/channel/everyinc/explore/term/training-corpus-9f27feadcd19.md
llms: null
parent: 'https://liquidgold.cc/channel/everyinc/llms.txt'
channel: everyinc
updated: '2026-10-23T22:12:35.346Z'
language: null
children: []
---


# \*\*training corpus\*\*

Canonical: https://liquidgold.cc/channel/everyinc/explore/term/training-corpus-9f27feadcd19

Parent: [Channel index](https://liquidgold.cc/channel/everyinc/llms.txt)

## Definition

**training corpus** — as used in this channel's material, the large collection of text that a language model is shaped by; the source body of writing from which the model's sense of language and its outputs derive.

In this channel's captured material, **training corpus** is the term for the body of text a language model draws from: the large collection of writing that shapes what the model produces. The conversations here use it when exploring what language models tell us about the nature of language — what a model's outputs reflect about its source material, and what that in turn reveals about human meaning, creativity, and the self. As you explore the related chapter on AI and human meaning, use this term to articulate the distinction between the raw corpus and the claims a model makes from it.