INDEX

Explanations

chess game context

The neuron fires on terminology related to chess problem composition (e.g. words like “composed,” “problems,” “composition,” “problemist”).

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

 then

-1.77

</i>

-1.44

who

-1.39

 into

-1.38

ﻷ

-1.37

ArrayAdapter

-1.36

}$,

-1.35

<td>

-1.34

šča

-1.34

it

-1.33

POSITIVE LOGITS

気がする

1.53

żo

1.34

美味し

1.29

 всі

1.27

美味しく

1.26

 refuser

1.25

لباس

1.25

 inimigo

1.23

 hochwer

1.23

屨

1.22

Activations Density 0.039%