INDEX

Explanations

code structure keywords

This neuron fires on programming‐language identifiers—especially long, CamelCase names (e.g. class, type, and plugin names) in source code.

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

on

-1.55

and

-1.41

 first

-1.21

 before

-1.16

 with

-1.10

-1.07

at

-1.07

in

-1.06

or

-1.06

 epitom

-1.05

POSITIVE LOGITS

 figliu

1.30

significantly

1.25

钢

1.23

rfloor

1.21

 deixa

1.20

 олив

1.20

 sostuvo

1.16

rceil

1.16

Ingredienti

1.16

 certificados

1.16

Activations Density 0.087%