INDEX

Explanations

class/property declarations

The neuron fires on tokens that are capitalized identifiers (e.g. class‐ or field‐style names in code and config), i.e. words beginning with an uppercase letter.

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

 diferentes

-1.44

 porque

-1.34

},

-1.21

 innings

-1.20

 involving

-1.20

 già

-1.17

kyat

-1.15

of

-1.14

 nehmen

-1.14

 olika

-1.13

POSITIVE LOGITS

 сказала

1.13

 neigh

1.13

 than

1.11

 ravine

1.08

for

1.04

 then

1.01

 powinien

1.01

 fonctionnalités

1.00

掙

0.99

 регулярно

0.99

Activations Density 0.086%