Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 2012 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.7254. Min Act: -0.0001

Data Index: 1454312 (Python Code)

Max Activating Token Index: 96

Text #1

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.7014. Min Act: -0.0000

Data Index: 1385006 (Python Code)

Max Activating Token Index: 23

Text #2

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6717. Min Act: -0.0000

Data Index: 591349 (C4 (Web Text))

Max Activating Token Index: 596

Text #3

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6287. Min Act: -0.0000

Data Index: 1516089 (Python Code)

Max Activating Token Index: 165

Text #4

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6199. Min Act: -0.0000

Data Index: 1366188 (Python Code)

Max Activating Token Index: 29

Text #5

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6349. Min Act: -0.0001

Data Index: 254520 (C4 (Web Text))

Max Activating Token Index: 375

Text #6

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6205. Min Act: -0.0001

Data Index: 510360 (C4 (Web Text))

Max Activating Token Index: 492

Text #7

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6044. Min Act: -0.0001

Data Index: 381608 (C4 (Web Text))

Max Activating Token Index: 356

Text #8

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.5951. Min Act: -0.0001

Data Index: 1621582 (Python Code)

Max Activating Token Index: 64

Text #9

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.4172. Min Act: -0.0000

Data Index: 760872 (C4 (Web Text))

Max Activating Token Index: 925

Text #10

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6080. Min Act: -0.0000

Data Index: 1357854 (C4 (Web Text))

Max Activating Token Index: 588

Text #11

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.5658. Min Act: -0.0000

Data Index: 1585791 (Python Code)

Max Activating Token Index: 25

Text #12

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.5844. Min Act: -0.0000

Data Index: 1434792 (Python Code)

Max Activating Token Index: 118

Text #13

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.5697. Min Act: -0.0001

Data Index: 1516325 (Python Code)

Max Activating Token Index: 71

Text #14

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.5785. Min Act: -0.0000

Data Index: 330388 (C4 (Web Text))

Max Activating Token Index: 459

Text #15

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.5406. Min Act: -0.0001

Data Index: 669330 (C4 (Web Text))

Max Activating Token Index: 689

Text #16

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6114. Min Act: -0.0001

Data Index: 196520 (C4 (Web Text))

Max Activating Token Index: 775

Text #17

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.4570. Min Act: -0.0001

Data Index: 545662 (C4 (Web Text))

Max Activating Token Index: 932

Text #18

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6043. Min Act: -0.0001

Data Index: 585955 (C4 (Web Text))

Max Activating Token Index: 264

Text #19

Max Range: 0.7254. Min Range: -0.7254

Max Act: 0.6285. Min Act: -0.0001

Data Index: 406808 (C4 (Web Text))

Max Activating Token Index: 786