INDEX

Explanations

testing and experimentation

The neuron strongly activates on occurrences of “test,” “tested,” “testing,” and other forms of that root.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

ana

1.28

1.26

দা

1.17

1.14

 handwriting

1.08

ul

1.05

vat

1.05

ed

1.04

av

1.01

POSITIVE LOGITS

试验

1.57

 nghiệm

1.52

IamPermissions

1.46

テスト

1.38

 TESTING

1.38

 测试

1.37

测试

1.31

 қили

1.27

 TESTS

1.27

Testing

1.27

Activations Density 0.131%