INDEX

Explanations

places and names

The neuron selectively responds to capitalized proper names and titles (i.e. named entities such as places, story or film titles, character names, etc.).

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

-1.46

-1.36

-1.24

-1.23

-1.22

を

-1.22

-1.18

どちらも

-1.16

分鐘

-1.16

が

-1.15

POSITIVE LOGITS

 their

1.87

 where

1.69

 From

1.29

by

1.29

 during

1.28

 Only

1.24

 dejan

1.22

 senere

1.17

 During

1.16

 wzglę

1.14

Activations Density 0.029%