INDEX

Explanations

DATA; LOAD DATA LOCAL

The neuron fires on uppercase identifier‐style tokens that are immediately followed by numeric digits (e.g. “DATA120”, “DATA144”), i.e. macro–like names suffixed with numbers.

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

-2.41

ቍ

-2.33

was

-2.20

 כשה

-2.19

Ἦ

-2.14

iwisata

-2.13

 also

-2.06

튿

-2.05

琿

-2.05



-2.05

POSITIVE LOGITS

ſu

2.83

стреча

2.78

нутрен

2.61

новь

2.52

2.48

With

2.42

 Atención

2.41

DATA

2.41

Initially

2.39

ون

2.34

Activations Density 0.001%