INDEX

Explanations

prepositions and conjunctions

The neuron strongly activates on page or section titles (i.e. multi-word proper names/headings).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 Cette

-1.28

ִּ

-1.20

țion

-1.16

 бампер

-1.14

 créer

-1.10

ַּ

-1.09

ほんとに

-1.09

ֵּ

-1.05

 média

-1.05

 largely

-1.04

POSITIVE LOGITS

 from

1.63

 with

1.36

 before

1.22

on

1.21

ׇ

1.19

 after

1.16

 despite

1.16

 without

1.15

 would

1.12

 amidst

1.11

Activations Density 0.007%