INDEX

Explanations

comparisons with than

This neuron detects the comparative particle “than” (as used in comparative constructions).

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

 with

-1.59

-1.55

-1.45

 also

-1.42

in

-1.35

are

-1.30

 særlig

-1.29

正好

-1.27

has

-1.25

盥

-1.24

POSITIVE LOGITS

 than

1.70

riched

1.67

glades

1.40

ће

1.35

we

1.34

 soldats

1.34

 Gegenteil

1.34

 باست

1.34

ván

1.33

 tubos

1.33

Activations Density 0.029%