INDEX

Explanations

past tense verbs, action verbs

The neuron detects numeric expressions and associated time-or quantity-related tokens (e.g. numbers, nights, weeks, check-in times, durations) in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

⽼

-1.47

 better

-1.40

 concep

-1.39

who

-1.31

you

-1.30

 gebouw

-1.30

战队

-1.30

 всѣ

-1.27

及其

-1.26

所有

-1.22

POSITIVE LOGITS

 only

1.35

 rámci

1.28

newswire

1.25

this

1.24

Currently

1.20

 hanya

1.19

 sær

1.17

任せ

1.17

Just

1.14

 yılında

1.14

Activations Density 0.070%