INDEX

Explanations

PRC, ancillary, userid

The neuron activates on uppercase abbreviations and acronyms (contiguous capital‐letter tokens).

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

 mijn

-2.70

胳

-2.59

↵↵

-2.58

ሖ

-2.56

⩕

-2.53

 rave

-2.47

 безопасность

-2.44

„

-2.42

 dois

-2.41

说是

-2.39

POSITIVE LOGITS

蟥

3.13

儈

3.02

ൎ

2.78

 curtail

2.70

2.58

鈊

2.56

翾

2.53

 露天

2.50

 zusätzliche

2.50

ooooo

2.50

Activations Density 0.020%