INDEX

Explanations

code identifiers and abbreviations

The neuron fires on numeric literals (especially decimal numbers) wherever they appear.

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

</td>

-1.73

to

-1.71

Cevap

-1.70

 pero

-1.43

 dolayı

-1.41

 ellos

-1.37

Kemudian

-1.37

 Wartawan

-1.34

of

-1.34

၀

-1.33

POSITIVE LOGITS

we

1.89

鲞

1.81

 repuesto

1.73

iramente

1.71

tividade

1.63

 eignet

1.62

peka

1.62

’

1.57

assisted

1.55

を

1.55

Activations Density 0.032%