INDEX

Explanations

code structures and punctuation

The neuron is tuned to spotting standalone numeric literals (digits) in code or markup, such as case labels, index values, and numeric string tokens.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 начали

-1.50

了我

-1.35

了下

-1.31

the

-1.30

wzi

-1.30

ادگی

-1.29

几人

-1.27

そうで

-1.26

違います

-1.25

 unserem

-1.23

POSITIVE LOGITS

方法

1.59

 they

1.58

are

1.49

offices

1.34

 utenti

1.34

))))

1.31

就是

1.29

 baden

1.27

 waer

1.27

 叶子

1.27

Activations Density 0.001%