INDEX

Explanations

as long as I can remember

This neuron detects expressions of temporal duration comparisons (e.g. “as long as,” “longer than”).

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

 voltak

-1.81

脚下

-1.67

aimeJ

-1.62

社の

-1.58

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.56

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.56

 nowa

-1.53

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.52

 Saltar

-1.52

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.51

POSITIVE LOGITS

st

1.88

ou

1.84

li

1.80

ven

1.77

ta

1.75

in

1.74

1.72

ti

1.71

ints

1.69

go

1.68

Activations Density 0.007%