INDEX

Explanations

this work

This neuron fires on numeric tokens—especially digit sequences like years (e.g. “1995”).

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

тек

-1.17

を行いました

-1.15

americana

-1.09

 "'");

-1.06

 without

-1.05

IONAL

-1.04

 \%)$

-1.04

若干

-1.02

for

-1.01

>--}}

-1.01

POSITIVE LOGITS

 these

1.63

 this

1.38

 этого

1.30

 này

1.20

1.18

 когда

1.11

 españolas

1.10

</strong>

1.09

 entraîner

1.09

ꈍ

1.08

Activations Density 0.301%