Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 516 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1674. Min Act: -0.0001

Data Index: 429785 (C4 (Web Text))

Max Activating Token Index: 716

Text #1

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1508. Min Act: -0.0002

Data Index: 1020897 (C4 (Web Text))

Max Activating Token Index: 406

Text #2

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1676. Min Act: -0.0002

Data Index: 1021441 (C4 (Web Text))

Max Activating Token Index: 303

Text #3

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1600. Min Act: -0.0002

Data Index: 606794 (C4 (Web Text))

Max Activating Token Index: 792

Text #4

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1535. Min Act: -0.0002

Data Index: 188279 (C4 (Web Text))

Max Activating Token Index: 907

Text #5

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1420. Min Act: -0.0001

Data Index: 1018014 (C4 (Web Text))

Max Activating Token Index: 487

Text #6

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1523. Min Act: -0.0002

Data Index: 1291024 (C4 (Web Text))

Max Activating Token Index: 982

Text #7

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1414. Min Act: -0.0002

Data Index: 1034122 (C4 (Web Text))

Max Activating Token Index: 114

Text #8

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1396. Min Act: -0.0001

Data Index: 1488690 (Python Code)

Max Activating Token Index: 150

Text #9

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1450. Min Act: -0.0001

Data Index: 994410 (C4 (Web Text))

Max Activating Token Index: 984

Text #10

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1285. Min Act: -0.0002

Data Index: 623014 (C4 (Web Text))

Max Activating Token Index: 573

Text #11

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1327. Min Act: -0.0002

Data Index: 1139758 (C4 (Web Text))

Max Activating Token Index: 153

Text #12

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1310. Min Act: -0.0002

Data Index: 314564 (C4 (Web Text))

Max Activating Token Index: 394

Text #13

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1336. Min Act: -0.0001

Data Index: 65552 (C4 (Web Text))

Max Activating Token Index: 947

Text #14

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1200. Min Act: -0.0001

Data Index: 1461133 (Python Code)

Max Activating Token Index: 533

Text #15

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1331. Min Act: -0.0002

Data Index: 96107 (C4 (Web Text))

Max Activating Token Index: 2

Text #16

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1418. Min Act: -0.0001

Data Index: 868052 (C4 (Web Text))

Max Activating Token Index: 1003

Text #17

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1256. Min Act: -0.0002

Data Index: 151828 (C4 (Web Text))

Max Activating Token Index: 507

Text #18

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1347. Min Act: -0.0001

Data Index: 775073 (C4 (Web Text))

Max Activating Token Index: 530

Text #19

Max Range: 0.1676. Min Range: -0.1676

Max Act: 0.1288. Min Act: -0.0001

Data Index: 175639 (C4 (Web Text))

Max Activating Token Index: 30