INDEX

Explanations

needed to/for

The neuron specifically detects occurrences of the word “needed.”

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

 caucho

-1.77

 acabados

-1.76

 rayas

-1.76

 burbujas

-1.75

 réservoir

-1.73

えっ

-1.70

 collègues

-1.67

 professeurs

-1.67

 помогает

-1.66

 puits

-1.66

POSITIVE LOGITS

to

5.34

for

3.64

in

2.31

 when

2.00

 для

1.73

1.56

 because

1.55

as

1.53

事的

1.51

Not

1.45

Activations Density 0.028%