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

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

Neuron 3407 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.4307. Min Act: -0.0000

Data Index: 294413 (C4 (Web Text))

Max Activating Token Index: 760

Text #1

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.4091. Min Act: 0.0000

Data Index: 205430 (C4 (Web Text))

Max Activating Token Index: 31

Text #2

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.4159. Min Act: -0.0000

Data Index: 1020341 (C4 (Web Text))

Max Activating Token Index: 256

Text #3

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3975. Min Act: 0.0001

Data Index: 1253054 (C4 (Web Text))

Max Activating Token Index: 759

Text #4

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.4001. Min Act: -0.0000

Data Index: 1244333 (C4 (Web Text))

Max Activating Token Index: 797

Text #5

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3548. Min Act: 0.0000

Data Index: 699813 (C4 (Web Text))

Max Activating Token Index: 225

Text #6

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3496. Min Act: 0.0000

Data Index: 206854 (C4 (Web Text))

Max Activating Token Index: 84

Text #7

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3390. Min Act: -0.0000

Data Index: 145293 (C4 (Web Text))

Max Activating Token Index: 837

Text #8

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3674. Min Act: 0.0000

Data Index: 915407 (C4 (Web Text))

Max Activating Token Index: 930

Text #9

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3618. Min Act: -0.0000

Data Index: 529176 (C4 (Web Text))

Max Activating Token Index: 315

Text #10

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3613. Min Act: -0.0000

Data Index: 1550092 (Python Code)

Max Activating Token Index: 929

Text #11

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3563. Min Act: -0.0000

Data Index: 1293332 (C4 (Web Text))

Max Activating Token Index: 34

Text #12

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3426. Min Act: -0.0000

Data Index: 1021534 (C4 (Web Text))

Max Activating Token Index: 199

Text #13

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3402. Min Act: -0.0000

Data Index: 1380780 (Python Code)

Max Activating Token Index: 164

Text #14

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3613. Min Act: -0.0000

Data Index: 409563 (C4 (Web Text))

Max Activating Token Index: 624

Text #15

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3494. Min Act: -0.0000

Data Index: 74234 (C4 (Web Text))

Max Activating Token Index: 228

Text #16

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3338. Min Act: -0.0000

Data Index: 1343696 (C4 (Web Text))

Max Activating Token Index: 813

Text #17

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3348. Min Act: 0.0001

Data Index: 407397 (C4 (Web Text))

Max Activating Token Index: 328

Text #18

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3469. Min Act: -0.0000

Data Index: 907681 (C4 (Web Text))

Max Activating Token Index: 154

Text #19

Max Range: 0.4307. Min Range: -0.4307

Max Act: 0.3240. Min Act: -0.0000

Data Index: 852627 (C4 (Web Text))

Max Activating Token Index: 57