INDEX

Explanations

code, questions, or specific phrases

The neuron fires on programming‐language declaration tokens (e.g. EXTERN, double‐underscore names, “void,” “ngx_…”), i.e. markers of code prototypes or external function declarations.

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

 have

-1.27

 edgy

-1.23

in

-1.20

 atlik

-1.20

ši

-1.16

guitarra

-1.14

\\

-1.13

 this

-1.13

 eccentric

-1.13

人には

-1.09

POSITIVE LOGITS

踯

1.27

躥

1.25

two

1.24

generally

1.23

 firſt

1.22

 préparer

1.22

onaise

1.21

tones

1.21

sy

1.20

 praticamente

1.20

Activations Density 0.000%