INDEX

Explanations

identifying questions

The neuron strongly activates on sensory‐description terms—especially flavor and aroma descriptors—in tasting‐note style text.

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

ּוֹ

-1.70

וֹ

-1.64

 бампер

-1.50

 riserv

-1.47

 السياس

-1.45

 设置

-1.34

for

-1.32

 quieran

-1.25

 impegn

-1.25

ִּ

-1.25

POSITIVE LOGITS

1.37

 ویتامین

1.30

 блюдо

1.28

ἡ

1.20

umers

1.20

 پروفایل

1.20

naf

1.18

aucun

1.15

 绘画

1.15

ὅ

1.14

Activations Density 0.005%