INDEX

Explanations

initiatives and attempts

This neuron detects words that denote initiatives or attempts—action-oriented nouns like “initiative,” “move,” “attempt,” “effort,” and “decision.”

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

ي

-1.73

י

-1.56

چه

-1.49

Ŝ

-1.45

מס

-1.45

🏨

-1.43

 polecam

-1.39

ير

-1.38

७

-1.37

栳

-1.36

POSITIVE LOGITS

by

1.73

涪

1.52

quatch

1.44

쵸

1.42

 shouldn

1.41

adone

1.39

 perpetuate

1.38

rebbe

1.38

this

1.38

it

1.37

Activations Density 0.016%