INDEX

Explanations

alternatively, how to find

The neuron fires strongly on common English “function” words (especially negations and other high‐frequency stopwords like “not,” “that,” “’s,” “all,” “for,” etc.).

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

for

-1.95

 that

-1.79

 first

-1.67

 three

-1.66

 give

-1.63

two

-1.63

or

-1.63

 five

-1.60

one

-1.59

by

-1.55

POSITIVE LOGITS

 فيديو

1.42

 wondrous

1.40

 diámetro

1.38

 ADDITIONAL

1.37

artigen

1.35

vertreter

1.34

 .......

1.33

 grumpy

1.30

 áng

1.30

ophone

1.30

Activations Density 0.051%