Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 428 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0489. Min Act: -0.0001

Data Index: 1213491 (C4 (Web Text))

Max Activating Token Index: 763

Text #1

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0471. Min Act: -0.0002

Data Index: 37357 (C4 (Web Text))

Max Activating Token Index: 332

Text #2

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0406. Min Act: -0.0001

Data Index: 1168917 (C4 (Web Text))

Max Activating Token Index: 924

Text #3

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0407. Min Act: -0.0001

Data Index: 1029227 (C4 (Web Text))

Max Activating Token Index: 592

Text #4

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0380. Min Act: -0.0002

Data Index: 93282 (C4 (Web Text))

Max Activating Token Index: 643

Text #5

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0402. Min Act: -0.0002

Data Index: 283251 (C4 (Web Text))

Max Activating Token Index: 894

Text #6

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0381. Min Act: -0.0002

Data Index: 1296156 (C4 (Web Text))

Max Activating Token Index: 807

Text #7

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0359. Min Act: -0.0002

Data Index: 994596 (C4 (Web Text))

Max Activating Token Index: 773

Text #8

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0346. Min Act: -0.0001

Data Index: 1463913 (Python Code)

Max Activating Token Index: 624

Text #9

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0327. Min Act: -0.0001

Data Index: 358387 (C4 (Web Text))

Max Activating Token Index: 956

Text #10

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0325. Min Act: -0.0001

Data Index: 873430 (C4 (Web Text))

Max Activating Token Index: 838

Text #11

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0322. Min Act: -0.0001

Data Index: 1635542 (Python Code)

Max Activating Token Index: 873

Text #12

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0312. Min Act: -0.0002

Data Index: 895039 (C4 (Web Text))

Max Activating Token Index: 505

Text #13

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0322. Min Act: -0.0002

Data Index: 670461 (C4 (Web Text))

Max Activating Token Index: 475

Text #14

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0310. Min Act: -0.0002

Data Index: 650587 (C4 (Web Text))

Max Activating Token Index: 805

Text #15

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0305. Min Act: -0.0001

Data Index: 7670 (C4 (Web Text))

Max Activating Token Index: 755

Text #16

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0310. Min Act: -0.0001

Data Index: 1598756 (Python Code)

Max Activating Token Index: 946

Text #17

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0304. Min Act: -0.0002

Data Index: 841958 (C4 (Web Text))

Max Activating Token Index: 905

Text #18

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0300. Min Act: -0.0002

Data Index: 141114 (C4 (Web Text))

Max Activating Token Index: 488

Text #19

Max Range: 0.0489. Min Range: -0.0489

Max Act: 0.0290. Min Act: -0.0002

Data Index: 39003 (C4 (Web Text))

Max Activating Token Index: 280