INDEX

Explanations

first time units

The neuron activates on numeric expressions of time—durations, ordinal “first”/“two,” and specific temporal spans like weeks, days, years, or nights.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 different

-1.67

Different

-1.43

 übernahm

-1.37

 hectáreas

-1.29

 dnes

-1.27

 الآن

-1.27

 waarmee

-1.27

 Different

-1.25

 malheure

-1.23

 entwickelte

-1.22

POSITIVE LOGITS

of

5.31

ของ

1.49

in

1.48

 της

1.38

 nawet

1.34

缡

1.23

anggung

1.22

 artig

1.22

 when

1.21

 there

1.17

Activations Density 0.021%