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

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

Neuron 4437 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: 0.7777. Min Range: -0.7777

Max Act: 0.7777. Min Act: -0.0001

Data Index: 1515143 (Python Code)

Max Activating Token Index: 107

Text #1

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6687. Min Act: -0.0001

Data Index: 860563 (C4 (Web Text))

Max Activating Token Index: 4

Text #2

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6663. Min Act: -0.0001

Data Index: 1509854 (Python Code)

Max Activating Token Index: 16

Text #3

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6286. Min Act: -0.0001

Data Index: 515792 (C4 (Web Text))

Max Activating Token Index: 162

Text #4

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6301. Min Act: -0.0001

Data Index: 1521677 (Python Code)

Max Activating Token Index: 259

Text #5

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6623. Min Act: -0.0001

Data Index: 578931 (C4 (Web Text))

Max Activating Token Index: 37

Text #6

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6557. Min Act: -0.0001

Data Index: 1317018 (C4 (Web Text))

Max Activating Token Index: 123

Text #7

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6465. Min Act: -0.0001

Data Index: 926864 (C4 (Web Text))

Max Activating Token Index: 86

Text #8

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6309. Min Act: -0.0001

Data Index: 528352 (C4 (Web Text))

Max Activating Token Index: 125

Text #9

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6251. Min Act: -0.0001

Data Index: 381217 (C4 (Web Text))

Max Activating Token Index: 41

Text #10

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.5556. Min Act: -0.0001

Data Index: 1596049 (Python Code)

Max Activating Token Index: 533

Text #11

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.5768. Min Act: -0.0001

Data Index: 83807 (C4 (Web Text))

Max Activating Token Index: 194

Text #12

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6089. Min Act: -0.0001

Data Index: 1023733 (C4 (Web Text))

Max Activating Token Index: 52

Text #13

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.5993. Min Act: -0.0001

Data Index: 1190636 (C4 (Web Text))

Max Activating Token Index: 105

Text #14

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.5806. Min Act: -0.0001

Data Index: 1431095 (Python Code)

Max Activating Token Index: 27

Text #15

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6000. Min Act: -0.0001

Data Index: 1108299 (C4 (Web Text))

Max Activating Token Index: 409

Text #16

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.5855. Min Act: -0.0001

Data Index: 1008901 (C4 (Web Text))

Max Activating Token Index: 265

Text #17

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.6084. Min Act: -0.0001

Data Index: 1486212 (Python Code)

Max Activating Token Index: 450

Text #18

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.5859. Min Act: -0.0001

Data Index: 379204 (C4 (Web Text))

Max Activating Token Index: 400

Text #19

Max Range: 0.7777. Min Range: -0.7777

Max Act: 0.5484. Min Act: -0.0001

Data Index: 846609 (C4 (Web Text))

Max Activating Token Index: 405