INDEX

Explanations

The neuron strongly activates on tokens containing apostrophes (e.g. contractions or possessives).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Contents

0.67

BEST

0.64

Best

0.63

‫

0.59

·

0.58

################

0.57

Chapter

0.57

Section

0.56

CHAPTER

0.56

////////////////

0.56

POSITIVE LOGITS

 inférieurs

0.56

 notValid

0.55

<unused2165>

0.53

(${

0.52

 tenere

0.51

 inférieures

0.51

 pomiesz

0.51

 foglie

0.50

 zadr

0.50

 Harus

0.50

Activations Density 0.026%