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

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

Neuron 640 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: 1.0098. Min Range: -1.0098

Max Act: 0.9853. Min Act: -0.0000

Data Index: 1045158 (C4 (Web Text))

Max Activating Token Index: 946

Text #1

Max Range: 1.0098. Min Range: -1.0098

Max Act: 1.0098. Min Act: -0.0000

Data Index: 558684 (C4 (Web Text))

Max Activating Token Index: 1001

Text #2

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.9806. Min Act: -0.0000

Data Index: 1195693 (C4 (Web Text))

Max Activating Token Index: 81

Text #3

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.9295. Min Act: -0.0000

Data Index: 43566 (C4 (Web Text))

Max Activating Token Index: 556

Text #4

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.8088. Min Act: -0.0000

Data Index: 321963 (C4 (Web Text))

Max Activating Token Index: 549

Text #5

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7964. Min Act: -0.0000

Data Index: 1245585 (C4 (Web Text))

Max Activating Token Index: 400

Text #6

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.8032. Min Act: -0.0000

Data Index: 1310198 (C4 (Web Text))

Max Activating Token Index: 987

Text #7

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.8063. Min Act: 0.0000

Data Index: 394556 (C4 (Web Text))

Max Activating Token Index: 930

Text #8

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7952. Min Act: 0.0000

Data Index: 1390579 (Python Code)

Max Activating Token Index: 846

Text #9

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7865. Min Act: 0.0000

Data Index: 692289 (C4 (Web Text))

Max Activating Token Index: 431

Text #10

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7855. Min Act: -0.0000

Data Index: 555305 (C4 (Web Text))

Max Activating Token Index: 792

Text #11

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7910. Min Act: -0.0000

Data Index: 1435782 (Python Code)

Max Activating Token Index: 1017

Text #12

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7199. Min Act: -0.0000

Data Index: 1462183 (Python Code)

Max Activating Token Index: 134

Text #13

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7440. Min Act: 0.0000

Data Index: 1576333 (Python Code)

Max Activating Token Index: 563

Text #14

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7317. Min Act: -0.0000

Data Index: 865771 (C4 (Web Text))

Max Activating Token Index: 395

Text #15

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7504. Min Act: -0.0000

Data Index: 1000906 (C4 (Web Text))

Max Activating Token Index: 941

Text #16

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7369. Min Act: -0.0000

Data Index: 1471485 (Python Code)

Max Activating Token Index: 954

Text #17

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7428. Min Act: -0.0000

Data Index: 1606073 (Python Code)

Max Activating Token Index: 261

Text #18

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7337. Min Act: 0.0000

Data Index: 1225563 (C4 (Web Text))

Max Activating Token Index: 750

Text #19

Max Range: 1.0098. Min Range: -1.0098

Max Act: 0.7391. Min Act: -0.0000

Data Index: 612616 (C4 (Web Text))

Max Activating Token Index: 880