INDEX

Explanations

from followed by prepositions

The neuron is selectively detecting the preposition “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

-1.76

</em>

-1.61

…”

-1.53

 said

-1.50

.”

-1.48

now

-1.48

 continued

-1.46

This

-1.41

</h2>

-1.41

POSITIVE LOGITS

the

1.75

legung

1.37

 empêche

1.36

 กับ

1.36

 Belajar

1.34

("").

1.32

felpa

1.32

гли

1.30

زندگی

1.28

 sorprendió

1.28

Activations Density 0.043%