INDEX

Explanations

emissions or equal

The neuron is detecting relatively rare or highly informative content words (i.e. tokens with high “IDF‐style” importance).

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

to

-1.55

 FBref

-1.34

 chay

-1.34

</table>

-1.31

 nuoc

-1.29

otheses

-1.29

</h2>

-1.29

]){

-1.28

 CAMERA

-1.27

 recommandations

-1.27

POSITIVE LOGITS

 including

1.61

or

1.56

 would

1.30

 veren

1.29

 especially

1.27

 when

1.27

at

1.26

esserts

1.24

 using

1.24

 through

1.24

Activations Density 0.547%