INDEX

Explanations

hints

The main thing this neuron does is detect occurrences of the word “hint” (including its morphological variants).

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

 réalisées

-2.55

 одном

-2.20

ارتمان

-2.16

 mutta

-2.14

 pero

-2.08

穢

-2.08

jenigen

-2.06

</

-2.06

żli

-2.03

 nhưng

-1.95

POSITIVE LOGITS

2.53

蜮

2.39

此事

2.33

 quedará

2.25

 fueran

2.25

 tivesse

2.17

快要

2.13

ком

2.08

锇

2.06

きは

2.03

Activations Density 0.004%