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

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

Neuron 1555 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: 0.0494. Min Range: -0.0494

Max Act: 0.0485. Min Act: -0.0004

Data Index: 715457 (C4 (Web Text))

Max Activating Token Index: 690

Text #1

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0490. Min Act: -0.0003

Data Index: 1607602 (Python Code)

Max Activating Token Index: 778

Text #2

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0494. Min Act: -0.0003

Data Index: 155838 (C4 (Web Text))

Max Activating Token Index: 606

Text #3

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0465. Min Act: -0.0004

Data Index: 708151 (C4 (Web Text))

Max Activating Token Index: 626

Text #4

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0411. Min Act: -0.0004

Data Index: 469102 (C4 (Web Text))

Max Activating Token Index: 393

Text #5

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0435. Min Act: -0.0003

Data Index: 297313 (C4 (Web Text))

Max Activating Token Index: 904

Text #6

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0421. Min Act: -0.0003

Data Index: 38465 (C4 (Web Text))

Max Activating Token Index: 846

Text #7

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0436. Min Act: -0.0004

Data Index: 550771 (C4 (Web Text))

Max Activating Token Index: 252

Text #8

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0394. Min Act: -0.0004

Data Index: 599185 (C4 (Web Text))

Max Activating Token Index: 490

Text #9

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0399. Min Act: -0.0003

Data Index: 535891 (C4 (Web Text))

Max Activating Token Index: 833

Text #10

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0408. Min Act: -0.0003

Data Index: 27643 (C4 (Web Text))

Max Activating Token Index: 718

Text #11

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0399. Min Act: -0.0003

Data Index: 1634624 (Python Code)

Max Activating Token Index: 726

Text #12

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0430. Min Act: -0.0003

Data Index: 1228703 (C4 (Web Text))

Max Activating Token Index: 698

Text #13

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0405. Min Act: -0.0003

Data Index: 1043545 (C4 (Web Text))

Max Activating Token Index: 500

Text #14

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0442. Min Act: -0.0004

Data Index: 1061608 (C4 (Web Text))

Max Activating Token Index: 923

Text #15

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0420. Min Act: -0.0003

Data Index: 793827 (C4 (Web Text))

Max Activating Token Index: 273

Text #16

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0427. Min Act: -0.0003

Data Index: 852742 (C4 (Web Text))

Max Activating Token Index: 664

Text #17

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0415. Min Act: -0.0003

Data Index: 466005 (C4 (Web Text))

Max Activating Token Index: 845

Text #18

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0421. Min Act: -0.0004

Data Index: 1187817 (C4 (Web Text))

Max Activating Token Index: 932

Text #19

Max Range: 0.0494. Min Range: -0.0494

Max Act: 0.0383. Min Act: -0.0004

Data Index: 1228981 (C4 (Web Text))

Max Activating Token Index: 522