INDEX

Explanations

do my best / job performance

This neuron detects praise constructions, especially “[Subject] does a/an [positive adjective] job” patterns (often with performer or actor names).

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

優惠

0.64

优惠

0.58

 healthier

0.55

 piac

0.52

 sweeter

0.52

ᓐ

0.52

राणिक

0.52

 healthiest

0.52

 صحت

0.51

 conveniente

0.51

POSITIVE LOGITS

job

1.45

 work

1.36

job

1.22

 trabajo

1.22

仕事

1.17

work

1.13

 pekerjaan

1.13

 работу

1.11

 práce

1.11

 práci

1.11

Activations Density 0.182%