INDEX

Explanations

ticket, send, cheated, to

This neuron fires on words used in property‐listing and directions contexts—i.e. terms that describe location, distance, and real-estate features.

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

いろいろ

0.86

さまざまな

0.86

そういう

0.82

 foodstuffs

0.79

 কাজকর্ম

0.79

WholeProgram

0.78

 таких

0.78

 takich

0.78

こうした

0.77

こういう

0.76

POSITIVE LOGITS

 FIVE

0.93

!!!!

0.92

!!!!!

0.89

!!!!!!

0.87

!!!

0.85

!!

0.85

 NEVER

0.83

!!!!!!!

0.83

0.82

 percentile

0.82

Activations Density 0.146%