Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

Dataset: 80% C4 (Web Text) and 20% Python Code

Neuron 2850 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 2.8251. Min Range: -2.8251

Max Act: 2.8251. Min Act: -0.0000

Data Index: 505104 (C4 (Web Text))

Max Activating Token Index: 836

Text #1

Max Range: 2.8251. Min Range: -2.8251

Max Act: 2.6482. Min Act: -0.0000

Data Index: 36181 (C4 (Web Text))

Max Activating Token Index: 56

Text #2

Max Range: 2.8251. Min Range: -2.8251

Max Act: 2.5748. Min Act: -0.0000

Data Index: 1036034 (C4 (Web Text))

Max Activating Token Index: 167

Text #3

Max Range: 2.8251. Min Range: -2.8251

Max Act: 2.4272. Min Act: -0.0000

Data Index: 150865 (C4 (Web Text))

Max Activating Token Index: 349

Text #4

Max Range: 2.8251. Min Range: -2.8251

Max Act: 2.4512. Min Act: -0.0000

Data Index: 1027283 (C4 (Web Text))

Max Activating Token Index: 377

Text #5

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.8546. Min Act: -0.0000

Data Index: 1449609 (Python Code)

Max Activating Token Index: 69

Text #6

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.8413. Min Act: -0.0000

Data Index: 1369325 (Python Code)

Max Activating Token Index: 441

Text #7

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.7510. Min Act: -0.0000

Data Index: 722015 (C4 (Web Text))

Max Activating Token Index: 199

Text #8

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.6289. Min Act: -0.0000

Data Index: 26478 (C4 (Web Text))

Max Activating Token Index: 493

Text #9

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.8074. Min Act: -0.0000

Data Index: 666378 (C4 (Web Text))

Max Activating Token Index: 843

Text #10

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.6967. Min Act: -0.0000

Data Index: 250275 (C4 (Web Text))

Max Activating Token Index: 510

Text #11

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.5938. Min Act: -0.0000

Data Index: 1477707 (Python Code)

Max Activating Token Index: 825

Text #12

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.5795. Min Act: -0.0000

Data Index: 1598076 (Python Code)

Max Activating Token Index: 414

Text #13

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.5626. Min Act: -0.0000

Data Index: 475514 (C4 (Web Text))

Max Activating Token Index: 38

Text #14

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.4893. Min Act: -0.0000

Data Index: 795106 (C4 (Web Text))

Max Activating Token Index: 670

Text #15

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.4870. Min Act: -0.0000

Data Index: 779670 (C4 (Web Text))

Max Activating Token Index: 845

Text #16

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.5611. Min Act: -0.0000

Data Index: 1390578 (Python Code)

Max Activating Token Index: 481

Text #17

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.5569. Min Act: -0.0000

Data Index: 1150099 (C4 (Web Text))

Max Activating Token Index: 460

Text #18

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.5629. Min Act: -0.0000

Data Index: 1000645 (C4 (Web Text))

Max Activating Token Index: 256

Text #19

Max Range: 2.8251. Min Range: -2.8251

Max Act: 1.4914. Min Act: -0.0000

Data Index: 1597333 (Python Code)

Max Activating Token Index: 443