INDEX

Explanations

could you ask

The neuron responds to tokens at sentence boundaries—especially the period “.” itself and the very next word—i.e. it marks the beginnings (and ends) of sentences.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

我們

3.88

 chúng

3.86

we

3.80

我们

3.75

 আমরা

3.70

我们在

3.64

我们也

3.59

เรา

3.58

私たちは

3.56

 우리는

3.53

POSITIVE LOGITS

私の

1.38

在我

1.25

 আমাকে

1.20

把我

1.18

对我

1.17

 mijn

1.16

 meiner

1.15

让我

1.13

 मेरे

1.13

 моих

1.12

Activations Density 0.384%