INDEX

Explanations

HTML body tags and code blocks

The neuron fires on programming code tokens—especially method or function names (often with parentheses) and code-style identifiers.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

तुम

0.57

ஞ்ஞான

0.54

 jelly

0.53

燃料

0.49

armonic

0.49

ulatory

0.48

均

0.48

PRESS

0.48

держи

0.47

 cirrh

0.47

POSITIVE LOGITS

0.96

{}

0.76

{})

0.71

{(

0.70

{[

0.69

{_

0.68

{};

0.67

{|

0.66

{$

0.66

{//

0.63

Activations Density 0.012%