Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 359 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.7802. Min Act: -0.0000

Data Index: 1440260 (Python Code)

Max Activating Token Index: 237

Text #1

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.5033. Min Act: -0.0000

Data Index: 385864 (C4 (Web Text))

Max Activating Token Index: 238

Text #2

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.5377. Min Act: -0.0000

Data Index: 2466 (C4 (Web Text))

Max Activating Token Index: 744

Text #3

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.5493. Min Act: -0.0000

Data Index: 994481 (C4 (Web Text))

Max Activating Token Index: 196

Text #4

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.5471. Min Act: -0.0000

Data Index: 926300 (C4 (Web Text))

Max Activating Token Index: 348

Text #5

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4822. Min Act: -0.0000

Data Index: 930366 (C4 (Web Text))

Max Activating Token Index: 610

Text #6

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4482. Min Act: -0.0000

Data Index: 794165 (C4 (Web Text))

Max Activating Token Index: 432

Text #7

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4595. Min Act: -0.0000

Data Index: 214571 (C4 (Web Text))

Max Activating Token Index: 895

Text #8

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4693. Min Act: -0.0000

Data Index: 1370953 (Python Code)

Max Activating Token Index: 323

Text #9

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4986. Min Act: -0.0000

Data Index: 1015827 (C4 (Web Text))

Max Activating Token Index: 856

Text #10

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4334. Min Act: -0.0000

Data Index: 337616 (C4 (Web Text))

Max Activating Token Index: 313

Text #11

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4479. Min Act: -0.0000

Data Index: 475971 (C4 (Web Text))

Max Activating Token Index: 873

Text #12

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4590. Min Act: -0.0000

Data Index: 1298949 (C4 (Web Text))

Max Activating Token Index: 678

Text #13

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4687. Min Act: -0.0000

Data Index: 1411917 (Python Code)

Max Activating Token Index: 448

Text #14

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4309. Min Act: -0.0000

Data Index: 1319053 (C4 (Web Text))

Max Activating Token Index: 301

Text #15

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4467. Min Act: -0.0000

Data Index: 1201088 (C4 (Web Text))

Max Activating Token Index: 703

Text #16

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4157. Min Act: -0.0000

Data Index: 1592195 (Python Code)

Max Activating Token Index: 493

Text #17

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4245. Min Act: -0.0000

Data Index: 789627 (C4 (Web Text))

Max Activating Token Index: 415

Text #18

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.4072. Min Act: -0.0000

Data Index: 811805 (C4 (Web Text))

Max Activating Token Index: 312

Text #19

Max Range: 0.7802. Min Range: -0.7802

Max Act: 0.3889. Min Act: -0.0000

Data Index: 161508 (C4 (Web Text))

Max Activating Token Index: 380