INDEX

Explanations

level, threshold, average, standard

The neuron detects nouns that denote quantitative benchmarks or standards (e.g. “level,” “threshold,” “average,” “standard,” “recommendation,” “scale,” “expectation”).

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

Tinh

-1.04

sendLogPayload

-1.02

AMES

-1.02

voyez

-1.02

دکتر

-0.99

kembangkan

-0.98

אח

-0.96

ھے

-0.95

}\

-0.95

 paysage

-0.94

POSITIVE LOGITS

of

1.82

for

1.27

by

1.11

 expectations

0.98

就像是

0.97

Даль

0.96

理由は

0.93

 bisogno

0.93

 teşekkür

0.93

 segíts

0.92

Activations Density 0.041%