INDEX

Explanations

was followed by the

The neuron strongly activates on past‐tense forms of the verb “to be” (especially “was” and “were”).

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

 currently

1.49

now

1.39

currently

1.32

現在

1.31

 বর্তমানে

1.26

 현재

1.25

 Currently

1.24

Currently

1.23

 actuellement

1.23

 obecnie

1.23

POSITIVE LOGITS

originally

1.28

 originally

1.22

 originalmente

1.16

 autrefois

1.16

 ursprünglich

1.15

 kemarin

1.09

 anteriormente

1.08

 gestern

1.06

 ontem

1.04

 damals

1.02

Activations Density 0.523%