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

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

Neuron 3973 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1380. Min Act: -0.0000

Data Index: 350789 (C4 (Web Text))

Max Activating Token Index: 845

Text #1

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1214. Min Act: -0.0000

Data Index: 1331825 (C4 (Web Text))

Max Activating Token Index: 806

Text #2

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1261. Min Act: -0.0000

Data Index: 1240405 (C4 (Web Text))

Max Activating Token Index: 154

Text #3

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1125. Min Act: -0.0000

Data Index: 388963 (C4 (Web Text))

Max Activating Token Index: 294

Text #4

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1135. Min Act: -0.0000

Data Index: 580218 (C4 (Web Text))

Max Activating Token Index: 344

Text #5

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1106. Min Act: -0.0000

Data Index: 1419023 (Python Code)

Max Activating Token Index: 93

Text #6

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1051. Min Act: -0.0000

Data Index: 602959 (C4 (Web Text))

Max Activating Token Index: 443

Text #7

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1034. Min Act: -0.0000

Data Index: 1236485 (C4 (Web Text))

Max Activating Token Index: 504

Text #8

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1090. Min Act: -0.0000

Data Index: 968014 (C4 (Web Text))

Max Activating Token Index: 823

Text #9

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1075. Min Act: -0.0000

Data Index: 401349 (C4 (Web Text))

Max Activating Token Index: 500

Text #10

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1003. Min Act: -0.0000

Data Index: 394214 (C4 (Web Text))

Max Activating Token Index: 450

Text #11

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1121. Min Act: -0.0000

Data Index: 1427217 (Python Code)

Max Activating Token Index: 783

Text #12

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.0996. Min Act: -0.0000

Data Index: 752292 (C4 (Web Text))

Max Activating Token Index: 490

Text #13

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1066. Min Act: -0.0000

Data Index: 753315 (C4 (Web Text))

Max Activating Token Index: 455

Text #14

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1036. Min Act: -0.0000

Data Index: 754692 (C4 (Web Text))

Max Activating Token Index: 820

Text #15

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1024. Min Act: -0.0000

Data Index: 224529 (C4 (Web Text))

Max Activating Token Index: 571

Text #16

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1014. Min Act: -0.0000

Data Index: 1622473 (Python Code)

Max Activating Token Index: 548

Text #17

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1063. Min Act: -0.0001

Data Index: 1330215 (C4 (Web Text))

Max Activating Token Index: 438

Text #18

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1047. Min Act: -0.0000

Data Index: 1280850 (C4 (Web Text))

Max Activating Token Index: 440

Text #19

Max Range: 0.1380. Min Range: -0.1380

Max Act: 0.1046. Min Act: -0.0000

Data Index: 442498 (C4 (Web Text))

Max Activating Token Index: 858