INDEX

Explanations

text, Text, edit, display

The neuron fires on user‐defined code identifiers—i.e. tokens that are part of variable, class, or resource names rather than language keywords.

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

räumen

-1.27

 such

-1.23

！

-1.17

Składniki

-1.16

 ミルク

-1.15

ŚĆ

-1.13

)}.

-1.09

Such

-1.09

-1.08

 didik

-1.07

POSITIVE LOGITS

 about

1.90

 before

1.40

 number

1.40

 only

1.39

 until

1.30

ばい

1.28

わりに

1.26

 \%}$

1.23

 from

1.21

 while

1.20

Activations Density 0.016%