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

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

Neuron 506 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: 0.0401. Min Range: -0.0401

Max Act: 0.0401. Min Act: -0.0002

Data Index: 1447934 (Python Code)

Max Activating Token Index: 626

Text #1

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0400. Min Act: -0.0002

Data Index: 762506 (C4 (Web Text))

Max Activating Token Index: 957

Text #2

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0382. Min Act: -0.0001

Data Index: 964584 (C4 (Web Text))

Max Activating Token Index: 485

Text #3

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0356. Min Act: -0.0002

Data Index: 1478389 (Python Code)

Max Activating Token Index: 508

Text #4

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0344. Min Act: -0.0001

Data Index: 803455 (C4 (Web Text))

Max Activating Token Index: 812

Text #5

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0349. Min Act: -0.0001

Data Index: 795481 (C4 (Web Text))

Max Activating Token Index: 747

Text #6

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0349. Min Act: -0.0002

Data Index: 1475562 (Python Code)

Max Activating Token Index: 800

Text #7

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0330. Min Act: -0.0002

Data Index: 1176445 (C4 (Web Text))

Max Activating Token Index: 55

Text #8

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0344. Min Act: -0.0002

Data Index: 746273 (C4 (Web Text))

Max Activating Token Index: 126

Text #9

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0326. Min Act: -0.0002

Data Index: 892892 (C4 (Web Text))

Max Activating Token Index: 216

Text #10

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0336. Min Act: -0.0001

Data Index: 768801 (C4 (Web Text))

Max Activating Token Index: 812

Text #11

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0323. Min Act: -0.0001

Data Index: 1167614 (C4 (Web Text))

Max Activating Token Index: 742

Text #12

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0334. Min Act: -0.0001

Data Index: 491198 (C4 (Web Text))

Max Activating Token Index: 965

Text #13

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0322. Min Act: -0.0002

Data Index: 1442240 (Python Code)

Max Activating Token Index: 515

Text #14

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0304. Min Act: -0.0002

Data Index: 1364698 (Python Code)

Max Activating Token Index: 562

Text #15

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0320. Min Act: -0.0001

Data Index: 630981 (C4 (Web Text))

Max Activating Token Index: 865

Text #16

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0318. Min Act: -0.0001

Data Index: 615401 (C4 (Web Text))

Max Activating Token Index: 510

Text #17

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0303. Min Act: -0.0002

Data Index: 388060 (C4 (Web Text))

Max Activating Token Index: 586

Text #18

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0311. Min Act: -0.0002

Data Index: 1289689 (C4 (Web Text))

Max Activating Token Index: 642

Text #19

Max Range: 0.0401. Min Range: -0.0401

Max Act: 0.0312. Min Act: -0.0002

Data Index: 373789 (C4 (Web Text))

Max Activating Token Index: 355