INDEX

Explanations

json url, sorted, packet, tts, hashed

The neuron strongly activates on programming identifiers—especially multi‐word variable or function names (e.g. tokens with underscores or mixed case).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 आल्सो

0.78

 ακόμα

0.59

 京都

0.56

 フォト

0.56

និច

0.56

 ακόμη

0.56

 पादुकोण

0.55

 フェ

0.54

 遊ん

0.54

 لأنه

0.54

POSITIVE LOGITS

1.38

="";

0.96

=[];

0.90

=[]

0.86

='';

0.85

=[

0.84

="

0.80

{};

0.79

={}

0.79

=''

0.75

Activations Density 0.005%