INDEX

Explanations

activities and processes

The neuron strongly activates on nouns or gerund forms referring to formal procedures or processes (e.g. maintenance, bidding, training, processing).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

努力

-1.04

 modificaciones

-0.98

 smer

-0.94

 plancher

-0.93

באנגלית

-0.92

 فقط

-0.91

 هیچ

-0.88

 منذ

-0.86

ADY

-0.85

發

-0.85

POSITIVE LOGITS

of

1.91

in

1.48

 done

1.32

 process

1.20

at

1.13

 facility

1.08

 mechanism

1.03

as

1.02

for

0.96

 through

0.94

Activations Density 0.051%