INDEX

Explanations

punctuation

The neuron detects snippets of regular‐expression syntax (e.g. backslashes, character‐class or group notation, quantifiers, etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

',()=>{

0.64

ແບບ

0.59

}></

0.57

ــ

0.57

ـــ

0.56

сила

0.55

horizontalLayout

0.55

ा

0.55

 উদ্দেশ্য

0.54

StructOf

0.53

POSITIVE LOGITS

4.00

3.03

*,

2.83

*;

2.63

*.

2.61

\*

2.61

*'

2.59

*',

2.58

*\

2.48

*",

2.48

Activations Density 0.087%