INDEX

Explanations

predicts and predictions

The neuron fires on terms introducing quantitative results or model predictions in scientific writing.

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

৭

-1.74

fromUtf

-1.71

ueroa

-1.70

</strong>

-1.66

 और

-1.66

-1.63

秶

-1.63

и

-1.62

("

-1.60

-1.59

POSITIVE LOGITS

shimano

1.73

皚

1.72

 acreditar

1.62

=[];

1.59

 概要

1.56

antics

1.56

 huidige

1.52

boite

1.52

 Jurí

1.51

 بايد

1.51

Activations Density 0.033%