INDEX

Explanations

limerick or poem requests

This neuron fires strongly on all‐caps headings or section titles (block‐leading uppercase words) in the text.

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

 powerhouse

0.69

or

0.67

或

0.66

 unleashed

0.65

 Kenyatta

0.65

 hesitant

0.64

 curvy

0.63

 cush

0.62

 embroiled

0.62

0.61

POSITIVE LOGITS

_.

0.95

*_

0.89

esterday

0.89

Hc

0.82

*.

0.82

lii

0.77

клары

0.77

(_

0.77

 interesantes

0.77

rfc

0.77

Activations Density 0.008%