INDEX

Explanations

correlate results map produce benefit associated returns

This neuron fires on verbs (and verb phrases) that express relationships or causation—e.g. “correlate,” “results in,” “map to,” “associated with,” “benefit from.”

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

 which

-1.54

۾

-1.30

which

-1.30

 mereka

-1.28

amento

-1.23

coder

-1.16

aneity

-1.16

ét

-1.13

thenia

-1.13

ción

-1.13

POSITIVE LOGITS

 hetzelfde

1.43

fmf

1.38

lxt

1.36

yyj

1.31

 propiedades

1.29

rmk

1.29

 takže

1.28

 reikia

1.27

Cdt

1.26

Einstellungen

1.25

Activations Density 0.239%