INDEX

Explanations

table separators

This neuron fires on long runs of non-word characters—e.g. sequences of underscores, question-marks, or other punctuation/gibberish placeholders.

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

 extravaganza

0.27

 vistas

0.24

 originates

0.24

 pastime

0.24

 bearers

0.23

 benevolence

0.23

करिता

0.23

 happenings

0.23

 trailbl

0.23

 objekt

0.23

POSITIVE LOGITS

0.27

 Yukarı

0.27

0.25

 Фі

0.25

滉

0.25

0.24

 Они

0.24

 Huyền

0.24

Activations Density 0.023%