INDEX

Explanations

becoming a state

The main thing this neuron does is detect occurrences of the verb “become” (in its various tenses/forms).

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

ма

1.76

५

1.48

ны

1.48

К

1.38

}$

1.37

ни

1.35

ती

1.31

ιώ

1.30

}$.

1.30

 construye

1.29

POSITIVE LOGITS

ಾದರೆ

1.16

 volna

1.16

ся

1.12

成为

1.10

了一

1.09

 dáng

1.09

et

1.07

了一个

1.06

なった

1.06

ದಲು

1.05

Activations Density 0.045%