INDEX

Explanations

questions starting with "why"

The neuron strongly activates on lines that start with an interrogative question word (e.g. “How,” “Which,” “What,” “Who,” “Why,” “Can”).

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

<bos>

-1.80

४

-1.39

mathrm

-1.25

цца

-1.20

ए

-1.20

ECTION

-1.18

 našich

-1.18

ी

-1.17

 גם

-1.13

ICTION

-1.13

POSITIVE LOGITS

 those

1.46

it

1.34

Those

1.31

 these

1.31

 September

1.30

 January

1.23

 November

1.20

及び

1.20

 september

1.17

 there

1.16

Activations Density 0.204%