INDEX

Explanations

future year references

This neuron is primarily detecting multi‐digit numeric tokens (e.g. year or season numbers).

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

 Peque

-1.79

!';

-1.78

^^^^^^^^

-1.75

',)

-1.74

ためには

-1.72

$*$

-1.69

뒀

-1.66

"..."

-1.63

--

-1.63

\",

-1.62

POSITIVE LOGITS

1.98

ஂ

1.81

↵↵

1.77

 vielleicht

1.63

 they

1.62

into

1.59

这里

1.59

 haha

1.57

 verwenden

1.56

 desarrollar

1.56

Activations Density 0.008%