INDEX

Explanations

Checklist for, Program Requirement:, Q:

This neuron responds strongly to hexadecimal byte values (hex digits like 0–9 and A–F) in the text.

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

 that

-1.35

 with

-1.34

 then

-1.24

in

-1.23

عارف

-1.22

at

-1.18

 olje

-1.18

 varme

-1.14

 without

-1.13

 aktivitet

-1.13

POSITIVE LOGITS

1.35

ar

1.23

er

1.19

 denselben

1.18

ad

1.16

ah

1.13

ue

1.13

一直在

1.09

al

1.07

Activations Density 0.002%