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

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

Neuron 3278 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.3229. Min Act: -0.0001

Data Index: 85576 (C4 (Web Text))

Max Activating Token Index: 653

Text #1

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.3087. Min Act: -0.0001

Data Index: 39741 (C4 (Web Text))

Max Activating Token Index: 993

Text #2

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.3233. Min Act: -0.0001

Data Index: 936621 (C4 (Web Text))

Max Activating Token Index: 182

Text #3

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.3059. Min Act: -0.0001

Data Index: 1326946 (C4 (Web Text))

Max Activating Token Index: 871

Text #4

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2930. Min Act: -0.0001

Data Index: 987577 (C4 (Web Text))

Max Activating Token Index: 228

Text #5

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.3026. Min Act: -0.0001

Data Index: 6762 (C4 (Web Text))

Max Activating Token Index: 514

Text #6

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.3083. Min Act: -0.0001

Data Index: 683826 (C4 (Web Text))

Max Activating Token Index: 238

Text #7

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2753. Min Act: -0.0001

Data Index: 363568 (C4 (Web Text))

Max Activating Token Index: 575

Text #8

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2680. Min Act: -0.0001

Data Index: 881073 (C4 (Web Text))

Max Activating Token Index: 976

Text #9

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2763. Min Act: -0.0001

Data Index: 704157 (C4 (Web Text))

Max Activating Token Index: 15

Text #10

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2751. Min Act: -0.0001

Data Index: 478801 (C4 (Web Text))

Max Activating Token Index: 1016

Text #11

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2766. Min Act: -0.0001

Data Index: 888534 (C4 (Web Text))

Max Activating Token Index: 764

Text #12

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2760. Min Act: -0.0000

Data Index: 372397 (C4 (Web Text))

Max Activating Token Index: 255

Text #13

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2533. Min Act: -0.0001

Data Index: 774722 (C4 (Web Text))

Max Activating Token Index: 184

Text #14

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2695. Min Act: -0.0001

Data Index: 1280271 (C4 (Web Text))

Max Activating Token Index: 122

Text #15

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2555. Min Act: -0.0001

Data Index: 937436 (C4 (Web Text))

Max Activating Token Index: 964

Text #16

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2517. Min Act: -0.0001

Data Index: 1584771 (Python Code)

Max Activating Token Index: 975

Text #17

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2436. Min Act: -0.0001

Data Index: 1300541 (C4 (Web Text))

Max Activating Token Index: 898

Text #18

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2407. Min Act: -0.0001

Data Index: 375460 (C4 (Web Text))

Max Activating Token Index: 318

Text #19

Max Range: 0.3233. Min Range: -0.3233

Max Act: 0.2352. Min Act: -0.0001

Data Index: 1178462 (C4 (Web Text))

Max Activating Token Index: 68