INDEX

Explanations

pull together data

The neuron fires on occurrences of the verb “pull” (and its inflected forms such as “pulls,” “pulling,” and “pull out”) in the text.

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

 neuen

-1.77

see

-1.65

to

-1.59

-1.57

at

-1.55

 его

-1.52

 when

-1.52

 want

-1.52

 find

-1.50

POSITIVE LOGITS

 ۱۷

1.73

逯

1.73

⢮

1.66

ridescent

1.61

 ۱۶

1.60

ης

1.59

ୌ

1.55

 blessé

1.55

 gezet

1.54

 freno

1.52

Activations Density 0.018%