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

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

Neuron 322 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: 0.0508. Min Range: -0.0508

Max Act: 0.0508. Min Act: -0.0003

Data Index: 935809 (C4 (Web Text))

Max Activating Token Index: 790

Text #1

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0372. Min Act: -0.0003

Data Index: 550556 (C4 (Web Text))

Max Activating Token Index: 369

Text #2

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0387. Min Act: -0.0003

Data Index: 949051 (C4 (Web Text))

Max Activating Token Index: 846

Text #3

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0317. Min Act: -0.0004

Data Index: 133559 (C4 (Web Text))

Max Activating Token Index: 756

Text #4

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0339. Min Act: -0.0002

Data Index: 904416 (C4 (Web Text))

Max Activating Token Index: 973

Text #5

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0325. Min Act: -0.0003

Data Index: 858475 (C4 (Web Text))

Max Activating Token Index: 593

Text #6

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0336. Min Act: -0.0003

Data Index: 1634429 (Python Code)

Max Activating Token Index: 736

Text #7

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0321. Min Act: -0.0003

Data Index: 1469632 (Python Code)

Max Activating Token Index: 868

Text #8

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0333. Min Act: -0.0003

Data Index: 1385531 (Python Code)

Max Activating Token Index: 665

Text #9

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0310. Min Act: -0.0003

Data Index: 723208 (C4 (Web Text))

Max Activating Token Index: 97

Text #10

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0327. Min Act: -0.0003

Data Index: 901516 (C4 (Web Text))

Max Activating Token Index: 714

Text #11

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0321. Min Act: -0.0002

Data Index: 431535 (C4 (Web Text))

Max Activating Token Index: 234

Text #12

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0311. Min Act: -0.0003

Data Index: 486949 (C4 (Web Text))

Max Activating Token Index: 925

Text #13

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0291. Min Act: -0.0003

Data Index: 1456717 (Python Code)

Max Activating Token Index: 947

Text #14

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0306. Min Act: -0.0003

Data Index: 720796 (C4 (Web Text))

Max Activating Token Index: 679

Text #15

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0318. Min Act: -0.0003

Data Index: 1457200 (Python Code)

Max Activating Token Index: 821

Text #16

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0303. Min Act: -0.0003

Data Index: 1448763 (Python Code)

Max Activating Token Index: 441

Text #17

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0294. Min Act: -0.0003

Data Index: 1425271 (Python Code)

Max Activating Token Index: 587

Text #18

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0267. Min Act: -0.0003

Data Index: 227233 (C4 (Web Text))

Max Activating Token Index: 777

Text #19

Max Range: 0.0508. Min Range: -0.0508

Max Act: 0.0301. Min Act: -0.0003

Data Index: 329288 (C4 (Web Text))

Max Activating Token Index: 944