INDEX

Explanations

inspect: or inspection item

The neuron fires on words that relate to examining or knowing something—e.g. “inspect,” “inspection,” and “familiar.”

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

ه

-2.78

-2.42



-2.38

齶

-2.25



-2.22

玬

-2.16

 PUSTAKA

-2.13

邸

-2.11

 evidenced

-2.08

 circunstancia

-2.08

POSITIVE LOGITS

咴

2.56

งิน

2.56

Зно

2.42

 Then

2.33

嗞

2.28

 количества

2.27

ally

2.25

鈇

2.22

 одним

2.19

Activations Density 0.000%