INDEX

Explanations

UIColor class checks

The neuron fires whenever it sees a Java class or type name (e.g. class literals, annotation‐ or API‐level type identifiers).

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

も

-0.82

參考文獻

-0.81

などが

-0.81

が

-0.81

つまり

-0.79

特に

-0.79

車は

-0.78

ных

-0.77

と

-0.77

vo

-0.76

POSITIVE LOGITS

class

1.96

 class

1.73

 ourselves

1.34

 themselves

1.09

 خود

1.08

 himself

1.05

themselves

1.04

 classes

1.02

 класса

1.02

 себя

1.01

Activations Density 0.005%