INDEX

Explanations

spelling and variety names

This neuron fires on standalone nouns denoting skills, tools, features, or qualifications—i.e. the keywords you’d see in a bullet-point list of required skills, product features, or technical specs.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 menu

0.67

 ovako

0.67

 certes

0.65

 tantôt

0.65

 takt

0.64

そういう

0.63

 tada

0.62

 ousted

0.61

 video

0.59

 manufacturers

0.59

POSITIVE LOGITS

៕

0.93

0.89

 உட்பட

0.88

.}$

0.84

 також

0.82

<eos>

0.81

ም።

0.81

 aswell

0.80

.。

0.79

 および

0.78

Activations Density 0.641%