INDEX

Explanations

spark industry consolidation

This neuron activates on words in title‐case headlines or section headings.

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

of

-2.30

and

-2.27

or

-2.02

 would

-1.95

 there

-1.91

but

-1.90

 during

-1.80

 could

-1.73

when

-1.68

 after

-1.67

POSITIVE LOGITS

 cuja

1.96

 assorted

1.88

艄

1.85

 conexiones

1.82

 rígida

1.81

his

1.80

 cujo

1.79

秾

1.78

 ancla

1.75

 disku

1.74

Activations Density 0.749%