INDEX

Explanations

variable names and parameters

This neuron fires on any token containing digits (numeric literals or identifiers with numeric suffixes).

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

of

-1.23

 strongly

-0.93

 viņ

-0.89

СТО

-0.89

unately

-0.88

 giudice

-0.86

 greatly

-0.85

 probably

-0.84

LETIN

-0.78

детский

-0.77

POSITIVE LOGITS

 indispens

0.88

 imbib

0.88

重い

0.87

 garg

0.85

iffel

0.84

 vorhanden

0.83

进了

0.82

Antes

0.80

another

0.79

 مشارکت

0.79

Activations Density 0.006%