INDEX

Explanations

recommendations or sequences

The neuron fires most strongly on common function words—especially the coordinating conjunction “and” (and to a lesser extent other stop-type tokens like “in,” “or,” commas, “not,” etc.).

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

 various

-1.27

deleteById

-1.11

 almost

-1.09

 caballeros

-1.04

fourths

-1.00

−/−

-0.99

Sejarah

-0.98

 VARIOUS

-0.97

 nearly

-0.97

 guste

-0.96

POSITIVE LOGITS

 обязательно

1.57

be

1.50

 preferably

1.49

to

1.34

 then

1.34

不要

1.16

if

1.15

 чтобы

1.14

 avoid

1.09

 тогда

1.06

Activations Density 0.027%