Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 4412 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 1.0472. Min Range: -1.0472

Max Act: 1.0472. Min Act: -0.0001

Data Index: 1368188 (Python Code)

Max Activating Token Index: 765

Text #1

Max Range: 1.0472. Min Range: -1.0472

Max Act: 1.0047. Min Act: -0.0001

Data Index: 1426340 (Python Code)

Max Activating Token Index: 577

Text #2

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.9151. Min Act: -0.0001

Data Index: 867338 (C4 (Web Text))

Max Activating Token Index: 641

Text #3

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.9542. Min Act: -0.0001

Data Index: 887535 (C4 (Web Text))

Max Activating Token Index: 40

Text #4

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8909. Min Act: -0.0001

Data Index: 1404749 (Python Code)

Max Activating Token Index: 295

Text #5

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.7786. Min Act: -0.0001

Data Index: 1653765 (Python Code)

Max Activating Token Index: 270

Text #6

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8986. Min Act: -0.0001

Data Index: 460860 (C4 (Web Text))

Max Activating Token Index: 143

Text #7

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8388. Min Act: -0.0001

Data Index: 1201423 (C4 (Web Text))

Max Activating Token Index: 49

Text #8

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.7335. Min Act: -0.0001

Data Index: 123642 (C4 (Web Text))

Max Activating Token Index: 648

Text #9

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8634. Min Act: -0.0001

Data Index: 1547809 (Python Code)

Max Activating Token Index: 211

Text #10

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8079. Min Act: -0.0001

Data Index: 1294237 (C4 (Web Text))

Max Activating Token Index: 737

Text #11

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8418. Min Act: -0.0001

Data Index: 146432 (C4 (Web Text))

Max Activating Token Index: 132

Text #12

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.7504. Min Act: -0.0001

Data Index: 1061974 (C4 (Web Text))

Max Activating Token Index: 304

Text #13

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.7767. Min Act: -0.0001

Data Index: 1590548 (Python Code)

Max Activating Token Index: 283

Text #14

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.7840. Min Act: -0.0001

Data Index: 1523138 (Python Code)

Max Activating Token Index: 411

Text #15

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8478. Min Act: -0.0001

Data Index: 721659 (C4 (Web Text))

Max Activating Token Index: 65

Text #16

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8183. Min Act: -0.0001

Data Index: 1541277 (Python Code)

Max Activating Token Index: 384

Text #17

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8459. Min Act: -0.0001

Data Index: 137905 (C4 (Web Text))

Max Activating Token Index: 439

Text #18

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.7898. Min Act: -0.0001

Data Index: 284674 (C4 (Web Text))

Max Activating Token Index: 573

Text #19

Max Range: 1.0472. Min Range: -1.0472

Max Act: 0.8327. Min Act: -0.0001

Data Index: 100449 (C4 (Web Text))

Max Activating Token Index: 57