INDEX

Explanations

research methods and studies

The neuron fires on nouns naming research methods or study types (e.g. “studies,” “analysis,” “method,” “technique,” “hypothesis”).

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

。

-1.45

мото

-1.40

DataSnapshot

-1.33

llten

-1.27

ziału

-1.27

كويكب

-1.25

ITIES

-1.21

ERS

-1.18

riert

-1.16

').

-1.14

POSITIVE LOGITS

 then

2.08

 different

1.48

 subsequently

1.40

 another

1.40

1.36

 from

1.33

 następnie

1.32

now

1.30

 during

1.30

 such

1.29

Activations Density 0.340%