INDEX

Explanations

the units digit, highest, prime

This neuron fires on the main thing being asked for in each math problem—i.e. the variable or entity whose value (digit, divisor, remainder, date of birth, etc.) the question is requesting.

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

 потребу

-1.06

 tanpa

-0.94

 into

-0.93

 REQUIREMENTS

-0.91

 they

-0.91

曰

-0.90

on

-0.88

 REQUIRE

-0.82

 ویکی‌پدیا

-0.79

 want

-0.79

POSITIVE LOGITS

要么

1.07

шных

0.96

ステン

0.93

czne

0.92

 original

0.91

Jika

0.89

我们会

0.88

 которы

0.88

ländische

0.88

mühle

0.86

Activations Density 0.001%