INDEX

Explanations

list indexing and data structures

The neuron fires on multi‐digit numeric literals (i.e. tokens representing larger numbers rather than single‐digit or 0–9 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

of

-1.60

by

-1.48

 during

-1.18

 even

-1.16

 either

-1.13

 while

-1.12

 colchón

-1.10

farol

-1.09

 used

-1.09

 using

-1.09

POSITIVE LOGITS

そして

1.32

 تاريخ

1.06

};

0.99

uarine

0.98

䛗

0.98



0.98

抡

0.97

Atl

0.97

/>

0.96

;">

0.96

Activations Density 0.034%