INDEX

Explanations

directory or input/output

The neuron spikes on specific named entities or brand‐style tokens (e.g. “Building,” “Facebook,” “Twitter”), i.e. it’s looking for proper names/brands.

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

}}$

0.87

}}{\

0.84

0.83

aurant

0.82

tut

0.80

aryana

0.80

以及

0.77

}}{

0.76

akili

0.76

POSITIVE LOGITS

 miejsc

0.92

 reminis

0.86

Sb

0.83

 landscaped

0.82

 cobbled

0.82

 hoisted

0.81

тые

0.79

 proveedores

0.78

 remastered

0.78

 قوي

0.77

Activations Density 0.000%