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

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

Neuron 1657 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: 1.0380. Min Range: -1.0380

Max Act: 1.0121. Min Act: -0.0001

Data Index: 64847 (C4 (Web Text))

Max Activating Token Index: 268

Text #1

Max Range: 1.0380. Min Range: -1.0380

Max Act: 1.0380. Min Act: -0.0001

Data Index: 1338935 (C4 (Web Text))

Max Activating Token Index: 120

Text #2

Max Range: 1.0380. Min Range: -1.0380

Max Act: 1.0051. Min Act: -0.0001

Data Index: 368070 (C4 (Web Text))

Max Activating Token Index: 18

Text #3

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.9966. Min Act: -0.0001

Data Index: 620522 (C4 (Web Text))

Max Activating Token Index: 355

Text #4

Max Range: 1.0380. Min Range: -1.0380

Max Act: 1.0255. Min Act: -0.0001

Data Index: 1240144 (C4 (Web Text))

Max Activating Token Index: 918

Text #5

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.9285. Min Act: -0.0001

Data Index: 1352867 (C4 (Web Text))

Max Activating Token Index: 312

Text #6

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.9282. Min Act: -0.0001

Data Index: 421432 (C4 (Web Text))

Max Activating Token Index: 545

Text #7

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.9400. Min Act: -0.0001

Data Index: 410816 (C4 (Web Text))

Max Activating Token Index: 605

Text #8

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.8519. Min Act: -0.0001

Data Index: 1250589 (C4 (Web Text))

Max Activating Token Index: 280

Text #9

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.9529. Min Act: -0.0001

Data Index: 871078 (C4 (Web Text))

Max Activating Token Index: 3

Text #10

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.9306. Min Act: -0.0001

Data Index: 266443 (C4 (Web Text))

Max Activating Token Index: 105

Text #11

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.8669. Min Act: -0.0001

Data Index: 118342 (C4 (Web Text))

Max Activating Token Index: 624

Text #12

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.9002. Min Act: -0.0001

Data Index: 201006 (C4 (Web Text))

Max Activating Token Index: 8

Text #13

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.8812. Min Act: -0.0001

Data Index: 1003559 (C4 (Web Text))

Max Activating Token Index: 235

Text #14

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.8440. Min Act: -0.0001

Data Index: 1389456 (Python Code)

Max Activating Token Index: 144

Text #15

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.8706. Min Act: -0.0001

Data Index: 798791 (C4 (Web Text))

Max Activating Token Index: 529

Text #16

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.8950. Min Act: -0.0001

Data Index: 461357 (C4 (Web Text))

Max Activating Token Index: 457

Text #17

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.8680. Min Act: -0.0001

Data Index: 355598 (C4 (Web Text))

Max Activating Token Index: 67

Text #18

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.8275. Min Act: -0.0001

Data Index: 1013712 (C4 (Web Text))

Max Activating Token Index: 797

Text #19

Max Range: 1.0380. Min Range: -1.0380

Max Act: 0.8587. Min Act: -0.0001

Data Index: 232931 (C4 (Web Text))

Max Activating Token Index: 13