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

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

Neuron 791 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: 0.1280. Min Range: -0.1280

Max Act: 0.1164. Min Act: -0.0001

Data Index: 347851 (C4 (Web Text))

Max Activating Token Index: 472

Text #1

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.1280. Min Act: -0.0001

Data Index: 321950 (C4 (Web Text))

Max Activating Token Index: 292

Text #2

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.1117. Min Act: -0.0001

Data Index: 565486 (C4 (Web Text))

Max Activating Token Index: 728

Text #3

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.1062. Min Act: -0.0001

Data Index: 906079 (C4 (Web Text))

Max Activating Token Index: 271

Text #4

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.1162. Min Act: -0.0001

Data Index: 1557881 (Python Code)

Max Activating Token Index: 96

Text #5

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.1043. Min Act: -0.0001

Data Index: 1000725 (C4 (Web Text))

Max Activating Token Index: 207

Text #6

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.1033. Min Act: -0.0001

Data Index: 919975 (C4 (Web Text))

Max Activating Token Index: 461

Text #7

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.1056. Min Act: -0.0001

Data Index: 988752 (C4 (Web Text))

Max Activating Token Index: 670

Text #8

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.1030. Min Act: -0.0001

Data Index: 1502839 (Python Code)

Max Activating Token Index: 78

Text #9

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0974. Min Act: -0.0001

Data Index: 1401247 (Python Code)

Max Activating Token Index: 295

Text #10

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.1057. Min Act: -0.0001

Data Index: 430631 (C4 (Web Text))

Max Activating Token Index: 172

Text #11

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0956. Min Act: -0.0001

Data Index: 211506 (C4 (Web Text))

Max Activating Token Index: 40

Text #12

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0935. Min Act: -0.0001

Data Index: 22745 (C4 (Web Text))

Max Activating Token Index: 980

Text #13

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0933. Min Act: -0.0001

Data Index: 1410462 (Python Code)

Max Activating Token Index: 348

Text #14

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0972. Min Act: -0.0001

Data Index: 1031609 (C4 (Web Text))

Max Activating Token Index: 201

Text #15

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0851. Min Act: -0.0001

Data Index: 1447617 (Python Code)

Max Activating Token Index: 690

Text #16

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0857. Min Act: -0.0001

Data Index: 931837 (C4 (Web Text))

Max Activating Token Index: 376

Text #17

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0876. Min Act: -0.0001

Data Index: 1120788 (C4 (Web Text))

Max Activating Token Index: 206

Text #18

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0906. Min Act: -0.0001

Data Index: 1650499 (Python Code)

Max Activating Token Index: 375

Text #19

Max Range: 0.1280. Min Range: -0.1280

Max Act: 0.0829. Min Act: -0.0001

Data Index: 382645 (C4 (Web Text))

Max Activating Token Index: 100