INDEX

Explanations

ensure something

The neuron activates on occurrences of the word “ensure” (typically in code identifiers or function names).

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

-1.73

 obtenidos

-1.73

and

-1.71

geſ

-1.69

We

-1.67

 attiv

-1.60

엿

-1.55

such

-1.54

那些

-1.54

栊

-1.52

POSITIVE LOGITS

ження

1.50

cinoma

1.50

)…

1.47

 there

1.46

 schützen

1.41

焌

1.41

 ‘‘

1.41

 zerst

1.38

ка

1.38

действие

1.36

Activations Density 0.006%