INDEX

Explanations

conjunctions and connecting words

acknowledging oddities or difficulties

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

′-

0.47

界的

0.44

′

0.43

सुद्धा

0.41

करों

0.40

⁽

0.40

 constituent

0.40

〈

0.38

^{\

0.38

 theorem

0.38

POSITIVE LOGITS

but

0.91

they

0.79

etc

0.78

which

0.74

there

0.73

we

0.72

this

0.72

 เอ่อ

0.71

hmm

0.71

you

0.70

Activations Density 0.084%