INDEX

Explanations

harder to

The neuron responds to comparative adjectives ending in “-er,” such as “harder.”

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

-2.13

邂

-2.06

ي

-1.98

-1.95

wiście

-1.95

 für

-1.92

τικής

-1.86

—

-1.84

rschein

-1.82

وضح

-1.79

POSITIVE LOGITS

the

2.73

 harder

2.39

蘩

2.34

CHREIB

2.28

 gröss

2.14

 offizielle

2.14

which

2.13

of

2.11

咴

2.09

 längere

2.02

Activations Density 0.002%