pronoun followed by had

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 불안

1.90

itado

1.87

EDULE

1.86

Kxc

1.86

iksaan

1.84

 rctx

1.80

 Fmat

1.79

킵

1.79

AsUpIndicator

1.75

ӡ

1.75

POSITIVE LOGITS

 exist

1.94

1.73

 exists

1.71

’

1.65

 existed

1.63

exist

1.56

had

1.53

1.52

 esist

1.52

 existe

1.50

Activations Density 0.247%

No Known Activations