INDEX

Explanations

first followed by a word

The neuron activates on occurrences of the word “first” (in titles or compound terms such as “Mobile First,” “Database First,” “safety first,” etc.).

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

\&

-1.53



-1.50

 huid

-1.48

 luxo

-1.47

 hukuman

-1.47

 EVERY

-1.45

чкой

-1.41

any

-1.41

</th>

-1.41

 multifaceted

-1.40

POSITIVE LOGITS

The

1.95

 своих

1.68

че

1.67

 somewhat

1.65

你的

1.59

 encendido

1.59

becca

1.55

 decidedly

1.52

we

1.51

Activations Density 0.010%