INDEX

Explanations

ATT costs

The neuron activates on numeric-statistical expressions—phrases giving averages, counts, or “per”-style metrics (e.g. “pictures per advert,” “average … has …”).

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

ی

0.80

cią

0.79

0.78

डाय

0.74

ور

0.73

ير

0.73

elems

0.73

Ricky

0.69

뎀

0.69

Will

0.68

POSITIVE LOGITS

 seasoning

0.95

 recombination

0.90

 revolver

0.85

 algal

0.85

 assortment

0.83

 ponder

0.83

ﻬ

0.82

 refrigeration

0.81

 recreation

0.80

 drugg

0.79

Activations Density 0.000%