INDEX

Explanations

network connection errors

The neuron is specialized to detect the word “connection” (and its close variants like “connections”).

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

</strong>

-1.80

-1.60

 maksimal

-1.56

 поча

-1.55

 hóspedes

-1.55

 menyen

-1.54

 mangiare

-1.53

 familières

-1.51

燃料

-1.50

and

-1.48

POSITIVE LOGITS

<bos>

2.03

 оли

1.84

ଃ

1.74

仕方

1.66

企业的

1.63

 слегка

1.62

好みの

1.59

 crece

1.59



1.59

HAD

1.55

Activations Density 0.028%