INDEX

Explanations

code, questions, or answers

The neuron fires strongly on the literal “(” token, i.e. it detects opening parentheses (especially in code contexts such as function or constructor calls).

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

 both

-1.31

 their

-1.27

its

-1.23

 Fünf

-1.09

 during

-1.05

 Both

-1.03

ったのは

-1.03

 before

-1.01

érc

-1.00

cerpts

-1.00

POSITIVE LOGITS

 préférences

1.08

Agen

1.03

 därför

0.99

 astfel

0.97

 températures

0.96

reducible

0.93

 chiffres

0.91

 caractères

0.90

eration

0.89

 digitale

0.89

Activations Density 0.000%