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

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

Neuron 1636 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: 0.0496. Min Range: -0.0496

Max Act: 0.0477. Min Act: -0.0003

Data Index: 1209121 (C4 (Web Text))

Max Activating Token Index: 673

Text #1

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0485. Min Act: -0.0003

Data Index: 897211 (C4 (Web Text))

Max Activating Token Index: 656

Text #2

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0496. Min Act: -0.0003

Data Index: 960758 (C4 (Web Text))

Max Activating Token Index: 704

Text #3

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0460. Min Act: -0.0003

Data Index: 1066898 (C4 (Web Text))

Max Activating Token Index: 175

Text #4

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0441. Min Act: -0.0003

Data Index: 1651603 (Python Code)

Max Activating Token Index: 206

Text #5

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0458. Min Act: -0.0003

Data Index: 302373 (C4 (Web Text))

Max Activating Token Index: 36

Text #6

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0438. Min Act: -0.0003

Data Index: 311967 (C4 (Web Text))

Max Activating Token Index: 311

Text #7

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0421. Min Act: -0.0003

Data Index: 382571 (C4 (Web Text))

Max Activating Token Index: 22

Text #8

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0447. Min Act: -0.0003

Data Index: 907204 (C4 (Web Text))

Max Activating Token Index: 513

Text #9

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0442. Min Act: -0.0003

Data Index: 438042 (C4 (Web Text))

Max Activating Token Index: 220

Text #10

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0426. Min Act: -0.0003

Data Index: 739219 (C4 (Web Text))

Max Activating Token Index: 26

Text #11

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0419. Min Act: -0.0002

Data Index: 1445224 (Python Code)

Max Activating Token Index: 264

Text #12

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0410. Min Act: -0.0003

Data Index: 838682 (C4 (Web Text))

Max Activating Token Index: 17

Text #13

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0427. Min Act: -0.0003

Data Index: 1079785 (C4 (Web Text))

Max Activating Token Index: 811

Text #14

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0388. Min Act: -0.0003

Data Index: 803621 (C4 (Web Text))

Max Activating Token Index: 989

Text #15

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0417. Min Act: -0.0003

Data Index: 674237 (C4 (Web Text))

Max Activating Token Index: 16

Text #16

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0412. Min Act: -0.0003

Data Index: 413987 (C4 (Web Text))

Max Activating Token Index: 481

Text #17

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0422. Min Act: -0.0002

Data Index: 871278 (C4 (Web Text))

Max Activating Token Index: 756

Text #18

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0406. Min Act: -0.0003

Data Index: 421994 (C4 (Web Text))

Max Activating Token Index: 807

Text #19

Max Range: 0.0496. Min Range: -0.0496

Max Act: 0.0413. Min Act: -0.0003

Data Index: 561833 (C4 (Web Text))

Max Activating Token Index: 552