INDEX

Explanations

and followed by

The neuron flags longer, multi‐syllable content words—especially technical or scientific terms.

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



-1.51

 applau

-1.41

⚢

-1.36

カッコいい

-1.29

 téléph

-1.25



-1.25

回来了

-1.24

 花纹

-1.23

 siè

-1.23

 coute

-1.20

POSITIVE LOGITS

 that

1.66

 what

1.42

 this

1.34

to

1.33

of

1.32

all

1.28

 should

1.23

 make

1.23

 find

1.16

 That

1.16

Activations Density 0.181%