INDEX

Explanations

image display and handling

The neuron fires on mentions of images—especially the word “image” (and its translations) and related file‐format references like “jpg” or “png.”

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

</sup>

-1.15

','

-1.13

>();

-1.11

(":");

-1.10

}{

-1.09

-1.06

)}(\

-1.06

 traditionnels

-1.03

 Yellow

-1.01

忙碌

-1.01

POSITIVE LOGITS

 eenvou

1.19

ﻻ

1.17

 قوله

1.04

 pubblicazione

1.01

quedas

0.97

 gambar

0.96

 image

0.95

に限

0.95

Freitag

0.95

 sodio

0.95

Activations Density 0.062%