INDEX

Explanations

throw away, thrower, throwback

The neuron fires strongly on the substring “throw,” i.e. occurrences of the word “throw.”

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

-1.03

from

-0.91

増加

-0.88

 меня

-0.87

 болез

-0.86

fficio

-0.84

ровал

-0.84

 diplomat

-0.83

IS

-0.81

POSITIVE LOGITS

 Throw

1.55

Throw

1.43

away

1.41

 throw

1.22

throw

1.21

 Throws

1.20

 away

1.17

aways

1.15

throwing

1.09

 throwing

1.05

Activations Density 0.007%