INDEX

Explanations

Location:

This neuron fires on lines that specify “Location” followed by numeric coordinate or address values.

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

 rispondere

-1.78

 scrivere

-1.71

ませんが

-1.69

 perintah

-1.66

 Traditionally

-1.63

ܜ

-1.61

 faceva

-1.60

 procedere

-1.59

 mettere

-1.59

</h5>

-1.59

POSITIVE LOGITS

 okam

1.60

of

1.58

1.56

来た

1.55

ではある

1.52

 enorme

1.52

 gering

1.51

包子

1.51

1.48

也会

1.48

Activations Density 0.016%