INDEX

Explanations

, A, friends, Presentation, Q

The neuron fires on comma tokens (i.e. it’s detecting commas).

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

 göm

-1.37

 corrom

-1.30

corrência

-1.29

 fuis

-1.26

 soud

-1.24

 futebol

-1.23

 miniaturka

-1.23

 concerne

-1.18

 Qualquer

-1.16

 plais

-1.16

POSITIVE LOGITS

 will

1.27

十五

1.27

has

1.25

 still

1.19

 Some

1.13

 also

1.11

and

1.11

now

1.09

 This

1.07

 with

1.03

Activations Density 0.000%