INDEX

Explanations

quote marks and punctuation

The neuron fires on tokens marking the boundaries of quoted speech (i.e. opening or closing quotation‐mark characters).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 projectors

0.57

 shortcomings

0.57

 twigs

0.55

 exemptions

0.54

 takers

0.54

 rumours

0.52

 spies

0.52

στημα

0.52

 satte

0.52

 যাতে

0.51

POSITIVE LOGITS

↵

0.77

이란

0.76

Wikipedia

0.73

這

0.73

 라고

0.71

這

0.70

意思是

0.70

 বলিয়াই

0.69

이라는

0.66

这句话

0.66

Activations Density 0.274%