INDEX

Explanations

The neuron strongly fires on programming‐style code tokens—API calls, function names, punctuation and other syntax elements that look like code.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 असतात

0.56

 incendio

0.54

 기름

0.52

棄

0.51

が終わ

0.51

의

0.50

 trakcie

0.50

🦆

0.50

सामान्यीकृत

0.50

 derailed

0.50

POSITIVE LOGITS

#!/

1.05

import

0.95

 import

0.83

 void

0.79

 public

0.78

def

0.77

 typedef

0.73

 proguardFiles

0.73

 class

0.72

public

0.71

Activations Density 0.097%