INDEX

Explanations

package structure, execSQL, Month

The neuron is detecting HTML `<table>` tags (including their start and end markers).

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

</b>

-2.64

-2.36

 then

-2.33

ляции

-2.31

);

-2.30

laints

-2.20

my

-2.19

𓆉

-2.19

 praktischen

-2.09

🧌

-2.09

POSITIVE LOGITS

</h3>

2.95

of

2.55

大体

2.36

,“

2.27

only

2.27

were

2.25

2.22

 quot

2.20

旁邊

2.14

曁

2.14

Activations Density 0.002%