Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 1858 in Layer 5

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.5211. Min Act: -0.0001

Data Index: 449199 (C4 (Web Text))

Max Activating Token Index: 565

Text #1

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.5952. Min Act: -0.0001

Data Index: 344137 (C4 (Web Text))

Max Activating Token Index: 677

Text #2

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.4487. Min Act: -0.0001

Data Index: 1402555 (Python Code)

Max Activating Token Index: 126

Text #3

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.4643. Min Act: -0.0001

Data Index: 130792 (C4 (Web Text))

Max Activating Token Index: 424

Text #4

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.4740. Min Act: -0.0001

Data Index: 1284652 (C4 (Web Text))

Max Activating Token Index: 160

Text #5

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.4106. Min Act: -0.0001

Data Index: 1259803 (C4 (Web Text))

Max Activating Token Index: 49

Text #6

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.3830. Min Act: -0.0001

Data Index: 375600 (C4 (Web Text))

Max Activating Token Index: 958

Text #7

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.4045. Min Act: -0.0001

Data Index: 477059 (C4 (Web Text))

Max Activating Token Index: 78

Text #8

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.2875. Min Act: -0.0001

Data Index: 1238995 (C4 (Web Text))

Max Activating Token Index: 279

Text #9

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.3533. Min Act: -0.0001

Data Index: 1464889 (Python Code)

Max Activating Token Index: 810

Text #10

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.3197. Min Act: -0.0001

Data Index: 2859 (C4 (Web Text))

Max Activating Token Index: 354

Text #11

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.3259. Min Act: -0.0001

Data Index: 782445 (C4 (Web Text))

Max Activating Token Index: 394

Text #12

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.3396. Min Act: -0.0001

Data Index: 561240 (C4 (Web Text))

Max Activating Token Index: 252

Text #13

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.3301. Min Act: -0.0001

Data Index: 1191496 (C4 (Web Text))

Max Activating Token Index: 320

Text #14

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.3256. Min Act: -0.0001

Data Index: 150167 (C4 (Web Text))

Max Activating Token Index: 884

Text #15

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.3509. Min Act: -0.0001

Data Index: 406790 (C4 (Web Text))

Max Activating Token Index: 586

Text #16

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.2473. Min Act: -0.0001

Data Index: 344061 (C4 (Web Text))

Max Activating Token Index: 570

Text #17

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.2896. Min Act: -0.0001

Data Index: 1493867 (Python Code)

Max Activating Token Index: 638

Text #18

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.2978. Min Act: -0.0001

Data Index: 159904 (C4 (Web Text))

Max Activating Token Index: 328

Text #19

Max Range: 0.5952. Min Range: -0.5952

Max Act: 0.2793. Min Act: -0.0001

Data Index: 940665 (C4 (Web Text))

Max Activating Token Index: 158