INDEX

Explanations

escalators and escalation

The neuron detects the “escal” substring (i.e. prefixes of words like escalate, escalator, escalation, Escalante).

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

踭

-2.53

The

-2.39

"¿

-2.38

웡

-2.38

峮

-2.31

).

-2.22

 effluents

-2.19

 Therefore

-2.19

小麦

-2.16

櫪

-2.14

POSITIVE LOGITS

𓋼

2.53

2.39

of

2.36

at

2.22

しても

2.16

or

2.14

KommentareTeilen

2.09

 Дне

2.09

er

2.06

on

2.06

Activations Density 0.001%