INDEX

Explanations

gentle actions

The neuron fires on words (mostly adverbs/adjectives) that describe a gentle, soft, or sweeping manner (e.g. “gently,” “soft,” “sweeping”).

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

</strong>

-2.05

 BOTH

-1.59

皷

-1.48

BUT

-1.47

AND

-1.44

 BEFORE

-1.37

锨

-1.34

 Members

-1.33

(%

-1.31

 volando

-1.30

POSITIVE LOGITS

1.65

 gently

1.58

 müßte

1.57

你不

1.49

鱟

1.34

1.33

."

1.32

 softly

1.30

 gentle

1.28

Activations Density 0.010%