INDEX

Explanations

is verb + descriptor

The neuron activates on tokens that are standalone numeric values (digit sequences).

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

焢

-1.76

and

-1.63

äsident

-1.52

晚

-1.44

tempted

-1.42

thentic

-1.41

vernote

-1.41

ternos

-1.41

gucig

-1.40

ヮ

-1.40

POSITIVE LOGITS

2.06

になった

1.54

meanor

1.45

↵↵↵↵↵↵↵

1.38

↵

1.33

ms

1.32

↵↵↵

1.30

The

1.29

会不会

1.29

系の

1.28

Activations Density 0.183%