INDEX

Explanations

have a/an/any

The neuron strongly activates on occurrences of the verb “have” (including its past form “had”).

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

ট

1.19

ال

1.14

ีย

1.12

 그렇

1.11

 hurting

1.05

 striking

1.05

 becoming

1.02

á

1.01

に

1.01

ม

1.00

POSITIVE LOGITS

 возможность

1.55

 recours

1.46

 vært

1.31

 recourse

1.28

 možnost

1.28

 можливість

1.26

 koristi

1.20

 tendency

1.20

 foresight

1.20

difficulty

1.19

Activations Density 1.003%