INDEX

Explanations

not followed by exceptions

This neuron activates on negation words (e.g. “not,” “never,” “unless”) to detect negative or prohibitive contexts.

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

or

-1.80

 only

-1.55

not

-1.42

any

-1.37

by

-1.35

 very

-1.32

so

-1.30

 blatantly

-1.30

 terribly

-1.30

out

-1.20

POSITIVE LOGITS

wwww

1.18

 …..

1.17

 thomas

1.17

!!!!!!!

1.15

instead

1.14

ita

1.14

teo

1.13

$$$

1.12

 PROFESSIONAL

1.11

💡

1.10

Activations Density 0.054%