INDEX

Explanations

comparative qualities

The neuron responds strongly to descriptive, evaluative adjectives (i.e. words characterizing qualities, like “compact,” “responsive,” “popular,” etc.).

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

 enough

-2.05

Enough

-1.48

 sufficiently

-1.38

enough

-1.32

 Enough

-1.31

 very

-1.30

 genug

-1.30

 genoeg

-1.25

 sufficient

-1.25

 mantienen

-1.20

POSITIVE LOGITS

 than

4.25

than

2.44

 niż

2.25

Than

2.06

and

1.96

 THAN

1.85

THAN

1.83

 Than

1.80

 version

1.57

กว่า

1.25

Activations Density 0.055%