INDEX

Explanations

effect, role, functions, or efficacy

This neuron selectively activates on technical/scientific content words that commonly appear in section headings or study descriptors—e.g. effect, function(s), role, description, mechanism, efficacy—marking structural cues in research texts.

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

炻

-1.80

VERTISEMENT

-1.79



-1.77

 akrab

-1.71

icii

-1.66

旆

-1.66

 parfü

-1.65

 menggunakan

-1.63

 первая

-1.60

 bedien

-1.60

POSITIVE LOGITS

↵↵

2.03

");

1.91

1.90

1.85

1.84

”

1.81

1.77

</b>

1.76

.”

1.71

The

1.70

Activations Density 0.037%