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

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

Neuron 3351 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1285. Min Act: -0.0001

Data Index: 1332979 (C4 (Web Text))

Max Activating Token Index: 944

Text #1

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1226. Min Act: -0.0001

Data Index: 1381867 (Python Code)

Max Activating Token Index: 34

Text #2

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1168. Min Act: -0.0001

Data Index: 69885 (C4 (Web Text))

Max Activating Token Index: 621

Text #3

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1242. Min Act: -0.0001

Data Index: 1497705 (Python Code)

Max Activating Token Index: 39

Text #4

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1004. Min Act: -0.0001

Data Index: 72112 (C4 (Web Text))

Max Activating Token Index: 168

Text #5

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1257. Min Act: -0.0001

Data Index: 1279661 (C4 (Web Text))

Max Activating Token Index: 951

Text #6

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.0962. Min Act: -0.0001

Data Index: 503123 (C4 (Web Text))

Max Activating Token Index: 5

Text #7

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1216. Min Act: -0.0001

Data Index: 819606 (C4 (Web Text))

Max Activating Token Index: 474

Text #8

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1045. Min Act: -0.0001

Data Index: 570716 (C4 (Web Text))

Max Activating Token Index: 115

Text #9

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1145. Min Act: -0.0001

Data Index: 648762 (C4 (Web Text))

Max Activating Token Index: 382

Text #10

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.0967. Min Act: -0.0001

Data Index: 1242331 (C4 (Web Text))

Max Activating Token Index: 927

Text #11

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.0766. Min Act: -0.0001

Data Index: 1600417 (Python Code)

Max Activating Token Index: 586

Text #12

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.0893. Min Act: -0.0001

Data Index: 657527 (C4 (Web Text))

Max Activating Token Index: 371

Text #13

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1007. Min Act: -0.0001

Data Index: 525456 (C4 (Web Text))

Max Activating Token Index: 645

Text #14

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1055. Min Act: -0.0001

Data Index: 1143898 (C4 (Web Text))

Max Activating Token Index: 298

Text #15

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1077. Min Act: -0.0001

Data Index: 894016 (C4 (Web Text))

Max Activating Token Index: 27

Text #16

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1130. Min Act: -0.0001

Data Index: 1450266 (Python Code)

Max Activating Token Index: 51

Text #17

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1023. Min Act: -0.0001

Data Index: 31410 (C4 (Web Text))

Max Activating Token Index: 63

Text #18

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1036. Min Act: -0.0001

Data Index: 558235 (C4 (Web Text))

Max Activating Token Index: 950

Text #19

Max Range: 0.1285. Min Range: -0.1285

Max Act: 0.1050. Min Act: -0.0001

Data Index: 1646414 (Python Code)

Max Activating Token Index: 51