AI ALIGNMENT FORUM
Wikitags
AF

Glitch Tokens

Written by Raymond Arnold, CronoDAS last updated 18th Apr 2023

Glitch Tokens are tokens in a language model that cause anomalous output, such as SolidGoldMagikarp.

Posts tagged Glitch Tokens

2

142SolidGoldMagikarp (plus, prompt generation)

Jessica Rumbelow, mwatkins

2y

17

0

20SolidGoldMagikarp III: Glitch token archaeology

mwatkins, Jessica Rumbelow

2y

3

0

41Anomalous tokens reveal the original identities of Instruct models

janus, John David Pressman

2y

7

0

25SolidGoldMagikarp II: technical details and more recent findings

mwatkins, Jessica Rumbelow

2y

0

0

23SmartyHeaderCode: anomalous tokens for GPT3.5 and GPT-4

2y

1