INDEX

Explanations

/ paths

The neuron flags tokens typical of command‐line and file‐path syntax (slashes, command names, “cmd”, refresh endpoints, etc.).

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

 Also

-1.81

our

-1.48

 planas

-1.45

 ricas

-1.33

-1.29

 urbanas

-1.28

 rús

-1.27

We

-1.26

 vraiment

-1.26

 tomadas

-1.25

POSITIVE LOGITS

;,

1.68

 दोनों

1.59

íba

1.54

zień

1.44

;&

1.44

 разные

1.43



1.43

mathrm

1.43

에는

1.39

 Beide

1.38

Activations Density 0.030%