INDEX

Explanations

start of phrases

This neuron detects LaTeX control sequences, especially environment-begin commands (tokens starting with “\,” like “\begin{…}”).

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

any

-1.06

↵↵↵

-0.88

TheReal

-0.87

as

-0.80

 frequently

-0.80

get

-0.79

 still

-0.78

it

-0.78

 often

-0.77

 regarding

-0.75

POSITIVE LOGITS

 beginning

1.13

begin

1.09

beginning

1.07

 شروع

0.94

 balsam

0.93

starting

0.93

 começ

0.93

ウス

0.91

 começo

0.89

cupine

0.87

Activations Density 0.003%