INDEX

Explanations

is the highest common factor of

This neuron detects words and phrases that introduce or form math‐problem questions (e.g. “What,” “List,” “highest common,” “prime factors,” “units digit,” 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

 женская

-1.38

了一会儿

-1.28

消費者

-1.23

 ليس

-1.22

<0x00>

-1.20

世間

-1.18

 graces

-1.17

CONCLUSIONES

-1.15

 детства

-1.15

ladı

-1.13

POSITIVE LOGITS

 also

1.59

 will

1.42

↵

1.42

 ваших

1.42

</h5>

1.32

そして

1.30

 These

1.26

 these

1.24

²/

1.20

 "));

1.19

Activations Density 0.064%