INDEX

Explanations

Final Answer: The final answer is

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

cap

1.36

 deep

1.22

 production

1.18

Ben

1.18

 actual

1.17

 background

1.14

 collection

1.13

 type

1.13

 attack

1.12

POSITIVE LOGITS

zynarod

2.45

asaddo

2.34

<unused1531>

2.29

<unused1516>

2.29

<unused1524>

2.26

<unused1513>

2.26

<unused1520>

2.26

<unused1530>

2.25

<unused1291>

2.24

<unused1590>

2.23

Activations Density 0.034%