INDEX

Explanations

do, did, does verbs

The neuron fires on the literal description strings and feature/example names you pass to testing DSL calls (e.g. the quoted text in describe/it/context/feature blocks).

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

-1.01

かかり

-0.83

?;

-0.83

];

-0.81

怎么说

-0.81

なのだ

-0.79

 artık

-0.79

 brut

-0.78

なんですね

-0.78

üyor

-0.78

POSITIVE LOGITS

do

3.25

did

2.52

 does

2.50

do

1.37

 should

1.27

Does

1.24

 will

1.15

 را

1.13

 shall

1.06

Did

1.05

Activations Density 0.013%