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

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

Neuron 625 in Layer 6

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.8177. Min Act: -0.0000

Data Index: 649423 (C4 (Web Text))

Max Activating Token Index: 822

Text #1

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.8543. Min Act: -0.0000

Data Index: 993575 (C4 (Web Text))

Max Activating Token Index: 262

Text #2

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.8054. Min Act: -0.0000

Data Index: 977049 (C4 (Web Text))

Max Activating Token Index: 5

Text #3

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.7970. Min Act: -0.0000

Data Index: 938868 (C4 (Web Text))

Max Activating Token Index: 926

Text #4

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.7940. Min Act: -0.0000

Data Index: 216823 (C4 (Web Text))

Max Activating Token Index: 34

Text #5

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.7849. Min Act: -0.0000

Data Index: 257364 (C4 (Web Text))

Max Activating Token Index: 265

Text #6

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.7834. Min Act: -0.0000

Data Index: 520848 (C4 (Web Text))

Max Activating Token Index: 64

Text #7

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.7617. Min Act: -0.0000

Data Index: 506718 (C4 (Web Text))

Max Activating Token Index: 730

Text #8

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.7635. Min Act: -0.0000

Data Index: 1625770 (Python Code)

Max Activating Token Index: 74

Text #9

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.7227. Min Act: -0.0000

Data Index: 1253674 (C4 (Web Text))

Max Activating Token Index: 703

Text #10

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.6999. Min Act: -0.0000

Data Index: 822167 (C4 (Web Text))

Max Activating Token Index: 1004

Text #11

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.6860. Min Act: -0.0000

Data Index: 347242 (C4 (Web Text))

Max Activating Token Index: 696

Text #12

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.6638. Min Act: -0.0000

Data Index: 792270 (C4 (Web Text))

Max Activating Token Index: 254

Text #13

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.7236. Min Act: -0.0000

Data Index: 856616 (C4 (Web Text))

Max Activating Token Index: 214

Text #14

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.6669. Min Act: -0.0000

Data Index: 75106 (C4 (Web Text))

Max Activating Token Index: 284

Text #15

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.6680. Min Act: -0.0000

Data Index: 171250 (C4 (Web Text))

Max Activating Token Index: 148

Text #16

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.6859. Min Act: -0.0000

Data Index: 1457994 (Python Code)

Max Activating Token Index: 109

Text #17

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.6775. Min Act: -0.0000

Data Index: 1354316 (C4 (Web Text))

Max Activating Token Index: 897

Text #18

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.6772. Min Act: -0.0000

Data Index: 154713 (C4 (Web Text))

Max Activating Token Index: 645

Text #19

Max Range: 0.8543. Min Range: -0.8543

Max Act: 0.6709. Min Act: -0.0000

Data Index: 181237 (C4 (Web Text))

Max Activating Token Index: 34