INDEX

Explanations

first

The neuron consistently activates on occurrences of the token “first.”

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

↵↵

-2.28

栞

-2.02

and

-2.02

 empleos

-1.93

 until

-1.93

↵

-1.91

 جمله

-1.78

but

-1.78

 alrededores

-1.76

 mittlere

-1.74

POSITIVE LOGITS

ly

2.61

 Coupled

2.52

From

2.47

first

2.42

ization

2.41

="

2.36

from

2.36

FIRST

2.34

2.33

which

2.33

Activations Density 0.007%