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

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

Neuron 4383 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: 0.7567. Min Range: -0.7567

Max Act: 0.7567. Min Act: 0.0000

Data Index: 387519 (C4 (Web Text))

Max Activating Token Index: 20

Text #1

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.6117. Min Act: -0.0000

Data Index: 86951 (C4 (Web Text))

Max Activating Token Index: 925

Text #2

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.6283. Min Act: -0.0000

Data Index: 1225586 (C4 (Web Text))

Max Activating Token Index: 104

Text #3

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5957. Min Act: -0.0000

Data Index: 1439715 (Python Code)

Max Activating Token Index: 560

Text #4

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5761. Min Act: -0.0000

Data Index: 1244929 (C4 (Web Text))

Max Activating Token Index: 224

Text #5

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5855. Min Act: -0.0000

Data Index: 728431 (C4 (Web Text))

Max Activating Token Index: 388

Text #6

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5459. Min Act: -0.0000

Data Index: 641555 (C4 (Web Text))

Max Activating Token Index: 147

Text #7

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.4945. Min Act: 0.0000

Data Index: 751447 (C4 (Web Text))

Max Activating Token Index: 777

Text #8

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.4380. Min Act: -0.0000

Data Index: 1495913 (Python Code)

Max Activating Token Index: 715

Text #9

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5170. Min Act: -0.0000

Data Index: 1455350 (Python Code)

Max Activating Token Index: 862

Text #10

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5186. Min Act: 0.0001

Data Index: 1045422 (C4 (Web Text))

Max Activating Token Index: 165

Text #11

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5417. Min Act: 0.0000

Data Index: 1171462 (C4 (Web Text))

Max Activating Token Index: 97

Text #12

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5428. Min Act: -0.0000

Data Index: 330721 (C4 (Web Text))

Max Activating Token Index: 38

Text #13

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5581. Min Act: -0.0000

Data Index: 877389 (C4 (Web Text))

Max Activating Token Index: 60

Text #14

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5558. Min Act: -0.0000

Data Index: 224561 (C4 (Web Text))

Max Activating Token Index: 785

Text #15

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5186. Min Act: 0.0000

Data Index: 95653 (C4 (Web Text))

Max Activating Token Index: 26

Text #16

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5344. Min Act: -0.0000

Data Index: 1569229 (Python Code)

Max Activating Token Index: 14

Text #17

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.5107. Min Act: -0.0000

Data Index: 472300 (C4 (Web Text))

Max Activating Token Index: 614

Text #18

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.4863. Min Act: 0.0000

Data Index: 417314 (C4 (Web Text))

Max Activating Token Index: 724

Text #19

Max Range: 0.7567. Min Range: -0.7567

Max Act: 0.4968. Min Act: -0.0000

Data Index: 1593848 (Python Code)

Max Activating Token Index: 417