INDEX

Explanations

references to the word "better" in context

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 good

-0.82

good

-0.78

Good

-0.78

 excellent

-0.73

 Good

-0.73

 GOOD

-0.72

GOOD

-0.64

excellent

-0.60

 EXCELL

-0.60

 gode

-0.59

POSITIVE LOGITS

 better

2.16

better

2.06

Better

1.98

 Better

1.93

 BETTER

1.72

 besser

1.48

 mieux

1.38

 meglio

1.34

 beter

1.34

 lepiej

1.27

Activations Density 0.034%