INDEX

Explanations

how can I

The neuron fires on occurrences of the modal “can” used in first‐person requests (e.g. “can I …”) indicating questions about what the speaker can do.

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

 mantienen

-1.35

ﺭ

-1.32

ޓ

-1.31

 最終

-1.27

rigkeit

-1.26

を探す

-1.25

 ͡°

-1.24

 匿名

-1.24

ídio

-1.23

bleiben

-1.23

POSITIVE LOGITS

.......

1.87

 .....

1.80

 ......

1.77

 ....

1.73

.....

1.71

.........

1.71

........

1.70

....

1.66

1.53

Activations Density 0.006%