Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 1583 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0502. Min Act: -0.0003

Data Index: 737722 (C4 (Web Text))

Max Activating Token Index: 588

Text #1

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0469. Min Act: -0.0003

Data Index: 1071924 (C4 (Web Text))

Max Activating Token Index: 375

Text #2

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0466. Min Act: -0.0002

Data Index: 1251617 (C4 (Web Text))

Max Activating Token Index: 935

Text #3

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0466. Min Act: -0.0003

Data Index: 502243 (C4 (Web Text))

Max Activating Token Index: 607

Text #4

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0404. Min Act: -0.0004

Data Index: 523007 (C4 (Web Text))

Max Activating Token Index: 460

Text #5

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0393. Min Act: -0.0002

Data Index: 511393 (C4 (Web Text))

Max Activating Token Index: 643

Text #6

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0388. Min Act: -0.0003

Data Index: 1049678 (C4 (Web Text))

Max Activating Token Index: 601

Text #7

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0404. Min Act: -0.0003

Data Index: 1080949 (C4 (Web Text))

Max Activating Token Index: 866

Text #8

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0401. Min Act: -0.0003

Data Index: 21896 (C4 (Web Text))

Max Activating Token Index: 660

Text #9

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0381. Min Act: -0.0003

Data Index: 435787 (C4 (Web Text))

Max Activating Token Index: 632

Text #10

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0411. Min Act: -0.0003

Data Index: 444215 (C4 (Web Text))

Max Activating Token Index: 811

Text #11

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0386. Min Act: -0.0003

Data Index: 258807 (C4 (Web Text))

Max Activating Token Index: 214

Text #12

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0377. Min Act: -0.0003

Data Index: 465072 (C4 (Web Text))

Max Activating Token Index: 625

Text #13

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0385. Min Act: -0.0003

Data Index: 2079 (C4 (Web Text))

Max Activating Token Index: 10

Text #14

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0364. Min Act: -0.0003

Data Index: 1418040 (Python Code)

Max Activating Token Index: 96

Text #15

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0376. Min Act: -0.0003

Data Index: 998065 (C4 (Web Text))

Max Activating Token Index: 652

Text #16

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0373. Min Act: -0.0003

Data Index: 604254 (C4 (Web Text))

Max Activating Token Index: 905

Text #17

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0391. Min Act: -0.0002

Data Index: 1305043 (C4 (Web Text))

Max Activating Token Index: 526

Text #18

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0370. Min Act: -0.0003

Data Index: 986946 (C4 (Web Text))

Max Activating Token Index: 652

Text #19

Max Range: 0.0502. Min Range: -0.0502

Max Act: 0.0368. Min Act: -0.0003

Data Index: 540042 (C4 (Web Text))

Max Activating Token Index: 748