INDEX

Explanations

pulling down or dragging

This neuron detects words describing physical dragging or pulling motions (especially downward).

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

 they

-2.08

новь

-2.02

 anún

-1.96

acuation

-1.84

 estadounid

-1.82

dik

-1.81

媄

-1.76

 jego

-1.73

 الجديدة

-1.69

 để

-1.66

POSITIVE LOGITS

</i>

2.03

</td>

2.00

</th>

1.74

幇

1.59

 slams

1.49

↵↵↵

1.47

运作

1.44

</b>

1.43

”;

1.43

</code>

1.42

Activations Density 0.002%