INDEX

Explanations

This neuron activates on the discourse‐marker phrase “the thing …” (e.g. “The thing is,” “The thing that…”) at the start of a clause.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

об

1.21

кре

1.05

ков

1.04

きた

1.02



1.02

DS

0.98

麒

0.98

υ

0.96

 quel

0.96

 cái

0.96

POSITIVE LOGITS

tions

1.59

لة

1.45

liness

1.38

 vibhav

1.35

wiście

1.31

ード

1.30

ികിത്സ

1.27

ग्वि

1.20

يًا

1.19

ratulations

1.18

Activations Density 0.075%