INDEX

Explanations

making or showing results

This neuron detects first-person help requests or instructional cues—phrases where the writer asks for guidance (e.g. “I’m trying to…,” “what did I miss,” “make sure to…,” “see…”).

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

 этого

-1.53

 tohoto

-1.48

 těchto

-1.48

of

-1.42

はこの

-1.41

 این

-1.39

 these

-1.38

 perciò

-1.38

was

-1.35

 dunque

-1.35

POSITIVE LOGITS

it

2.25

它的

1.47

 (;;)

1.37

 продаже

1.35

"]

1.30

 "..\..\

1.27

1.26

ш

1.25

1.24

А

1.22

Activations Density 0.201%