INDEX

Explanations

mathematical notation

This neuron fires on LaTeX/math formatting tokens (e.g. “$”, “^”, “Omega”, “partial”, “cases”, etc.)—i.e. it detects the presence of in‐line or display math syntax.

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

but

-1.23

}}=\

-0.94

心想

-0.89

지만

-0.88

and

-0.85

 selaku

-0.82

}}-\

-0.81

 menengah

-0.81

)}=\

-0.81

 rather

-0.80

POSITIVE LOGITS

,</

1.14

rawdź

1.02

>,</

1.02

 chistes

0.97

habitants

0.94

 SEÑ

0.94

Groetjes

0.93

 boire

0.92

少爷

0.91

(*)

0.91

Activations Density 0.006%