INDEX

Explanations

dates and range

The neuron is selectively responding to semantically “heavy” content words—i.e. keywords or topic-bearing terms—rather than common function words.

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

 while

-1.15

 sezonu

-1.10

paru

-1.02

cartier

-0.96

})(

-0.95

 означа

-0.94

 sizeof

-0.93

faktor

-0.92

ประก

-0.92

miami

-0.92

POSITIVE LOGITS

搅拌

1.09

ஂ

1.05

it

1.03

 faptul

1.01

 this

0.97

end

0.96

made

0.96

cas

0.96

AUF

0.95

欧式

0.95

Activations Density 0.001%