INDEX

Explanations

named places and brands

The neuron fires on multiword capitalized proper names (e.g. brand, place, or organization names).

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

擽

-2.00

つた

-1.98

鑁

-1.93

敺

-1.84

 moldura

-1.84

 zoude

-1.81

 oído

-1.80

 fiables

-1.77

 éte

-1.77

饷

-1.77

POSITIVE LOGITS

 also

2.20

",

2.17

in

2.17

and

2.16

 includes

1.80

This

1.77

not

1.74

."

1.73

や

1.67

of

1.65

Activations Density 0.122%