INDEX

Explanations

things that are difficult

This neuron specializes in spotting expressions of difficulty or challenge, particularly the adjective “difficult” (and related noun “difficulty”) as in “difficult to ….”

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

The

-2.17

acterísticas

-1.72

ﾆ

-1.71

ؿ

-1.52

曼

-1.48



-1.45

This

-1.44

ᨆ

-1.43

㊂

-1.43

 土産

-1.41

POSITIVE LOGITS

is

1.70

ılığı

1.67

 будет

1.66

 они

1.64

1.59

ließen

1.56

1.55

</b>

1.52

 הם

1.48

 принимает

1.44

Activations Density 0.004%