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

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

Neuron 2522 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: 1.4738. Min Range: -1.4738

Max Act: 1.4738. Min Act: -0.0001

Data Index: 455349 (C4 (Web Text))

Max Activating Token Index: 738

Text #1

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.3495. Min Act: -0.0001

Data Index: 116607 (C4 (Web Text))

Max Activating Token Index: 57

Text #2

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.4588. Min Act: -0.0001

Data Index: 511574 (C4 (Web Text))

Max Activating Token Index: 106

Text #3

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.3566. Min Act: -0.0001

Data Index: 466991 (C4 (Web Text))

Max Activating Token Index: 341

Text #4

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.3582. Min Act: -0.0001

Data Index: 345291 (C4 (Web Text))

Max Activating Token Index: 57

Text #5

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.3219. Min Act: -0.0001

Data Index: 1254707 (C4 (Web Text))

Max Activating Token Index: 202

Text #6

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.2432. Min Act: -0.0001

Data Index: 881983 (C4 (Web Text))

Max Activating Token Index: 123

Text #7

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.2449. Min Act: -0.0001

Data Index: 1342264 (C4 (Web Text))

Max Activating Token Index: 340

Text #8

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.2493. Min Act: -0.0001

Data Index: 1137883 (C4 (Web Text))

Max Activating Token Index: 732

Text #9

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.1717. Min Act: -0.0001

Data Index: 40364 (C4 (Web Text))

Max Activating Token Index: 819

Text #10

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.1097. Min Act: -0.0001

Data Index: 878842 (C4 (Web Text))

Max Activating Token Index: 851

Text #11

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.2306. Min Act: -0.0001

Data Index: 1013190 (C4 (Web Text))

Max Activating Token Index: 325

Text #12

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.0904. Min Act: -0.0001

Data Index: 824063 (C4 (Web Text))

Max Activating Token Index: 265

Text #13

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.1905. Min Act: -0.0001

Data Index: 1526268 (Python Code)

Max Activating Token Index: 593

Text #14

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.1160. Min Act: -0.0001

Data Index: 353648 (C4 (Web Text))

Max Activating Token Index: 914

Text #15

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.1276. Min Act: -0.0001

Data Index: 1143260 (C4 (Web Text))

Max Activating Token Index: 889

Text #16

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.0889. Min Act: -0.0001

Data Index: 487734 (C4 (Web Text))

Max Activating Token Index: 127

Text #17

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.0821. Min Act: -0.0001

Data Index: 4010 (C4 (Web Text))

Max Activating Token Index: 28

Text #18

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.0681. Min Act: -0.0001

Data Index: 615475 (C4 (Web Text))

Max Activating Token Index: 224

Text #19

Max Range: 1.4738. Min Range: -1.4738

Max Act: 1.1330. Min Act: -0.0001

Data Index: 240924 (C4 (Web Text))

Max Activating Token Index: 94