INDEX

Explanations

they are

The neuron selectively activates on third‐person plural references (e.g. “they,” “their,” “are”) and closely associated function words.

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

 behaupten

-1.27

segue

-1.16

 شما

-1.16

 multitude

-1.15

 shave

-1.14

 ilości

-1.13

 clinging

-1.10

 Avez

-1.10

Дети

-1.09

 cerámica

-1.09

POSITIVE LOGITS

 that

1.28

 exactamente

1.20

 probablemente

1.14

 another

1.13

 exatamente

1.12

ेशा

1.11

rased

1.05

などで

1.05

 begun

1.04

 here

1.03

Activations Density 0.185%