INDEX

Explanations

user interaction with controls

The neuron flags phrasing that describes user-driven UI actions or events (e.g. “when the user selects/clicks…”).

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

and

-1.49

was

-1.41

 business

-1.33

那些

-1.27

("

-1.27

 still

-1.27

 risk

-1.26

 quando

-1.24

all

-1.23

 when

-1.20

POSITIVE LOGITS

 there

1.56

 überprü

1.52

也不會

1.47

 vierten

1.45

 bleib

1.43



1.38

そのような

1.37

byshev

1.36

vœ

1.36

acknow

1.36

Activations Density 0.350%