INDEX

Explanations

Sol, followed by comma or Courts

This neuron is triggered by capitalized tokens, notably proper names, acronyms, and other uppercase or title‐case words.

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

caso

1.33

productos

1.31

rophages

1.28

编写

1.27

ంతరం

1.24

phones

1.23

aadhar

1.23

吗

1.22

carga

1.21

usermodel

1.20

POSITIVE LOGITS

 Officer

1.31

 Commander

1.30

1.28

1.27

 Emeritus

1.27

Off

1.25

1.23

1.22

 King

1.22

 Code

1.21

Activations Density 7.075%