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

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

Neuron 4713 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: 1.3782. Min Range: -1.3782

Max Act: 1.3782. Min Act: -0.0001

Data Index: 761953 (C4 (Web Text))

Max Activating Token Index: 488

Text #1

Max Range: 1.3782. Min Range: -1.3782

Max Act: 1.2877. Min Act: -0.0001

Data Index: 1182300 (C4 (Web Text))

Max Activating Token Index: 376

Text #2

Max Range: 1.3782. Min Range: -1.3782

Max Act: 1.2859. Min Act: -0.0001

Data Index: 115019 (C4 (Web Text))

Max Activating Token Index: 122

Text #3

Max Range: 1.3782. Min Range: -1.3782

Max Act: 1.1235. Min Act: -0.0001

Data Index: 461969 (C4 (Web Text))

Max Activating Token Index: 665

Text #4

Max Range: 1.3782. Min Range: -1.3782

Max Act: 1.1254. Min Act: -0.0001

Data Index: 95791 (C4 (Web Text))

Max Activating Token Index: 41

Text #5

Max Range: 1.3782. Min Range: -1.3782

Max Act: 1.1176. Min Act: -0.0001

Data Index: 351967 (C4 (Web Text))

Max Activating Token Index: 869

Text #6

Max Range: 1.3782. Min Range: -1.3782

Max Act: 1.0028. Min Act: -0.0001

Data Index: 1325057 (C4 (Web Text))

Max Activating Token Index: 816

Text #7

Max Range: 1.3782. Min Range: -1.3782

Max Act: 1.0671. Min Act: -0.0001

Data Index: 403263 (C4 (Web Text))

Max Activating Token Index: 276

Text #8

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9571. Min Act: -0.0001

Data Index: 33191 (C4 (Web Text))

Max Activating Token Index: 74

Text #9

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9575. Min Act: -0.0000

Data Index: 1458419 (Python Code)

Max Activating Token Index: 497

Text #10

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9555. Min Act: -0.0000

Data Index: 1542745 (Python Code)

Max Activating Token Index: 721

Text #11

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9155. Min Act: -0.0001

Data Index: 53858 (C4 (Web Text))

Max Activating Token Index: 322

Text #12

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9410. Min Act: -0.0001

Data Index: 1247877 (C4 (Web Text))

Max Activating Token Index: 120

Text #13

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9952. Min Act: -0.0001

Data Index: 1309826 (C4 (Web Text))

Max Activating Token Index: 1005

Text #14

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9245. Min Act: -0.0001

Data Index: 1570549 (Python Code)

Max Activating Token Index: 592

Text #15

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.8811. Min Act: -0.0001

Data Index: 120060 (C4 (Web Text))

Max Activating Token Index: 89

Text #16

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9661. Min Act: -0.0001

Data Index: 1166904 (C4 (Web Text))

Max Activating Token Index: 155

Text #17

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9245. Min Act: -0.0001

Data Index: 434085 (C4 (Web Text))

Max Activating Token Index: 476

Text #18

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9141. Min Act: -0.0000

Data Index: 1385488 (Python Code)

Max Activating Token Index: 643

Text #19

Max Range: 1.3782. Min Range: -1.3782

Max Act: 0.9105. Min Act: -0.0001

Data Index: 677602 (C4 (Web Text))

Max Activating Token Index: 263