INDEX

Explanations

the word "the" with a word before it and a high activation on the word directly before "the", also code that contains words relating to version control

say "the"

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

Negative Logits

RegressionTest

-0.76

the

-0.74

AddTagHelper

-0.66

borderFill

-0.63

(*(

-0.59

Ihe

-0.59

die

-0.59

 linkovi

-0.57

uttosto

-0.56

ihe

-0.56

POSITIVE LOGITS

de

0.81

0.74

Rè

0.55

т

0.54

SaveChangesAsync

0.52

0.51

0.48

">—

0.47

anément

0.47

Activations Density 1.423%