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

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

Neuron 4406 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: 0.1469. Min Range: -0.1469

Max Act: 0.1469. Min Act: -0.0001

Data Index: 810427 (C4 (Web Text))

Max Activating Token Index: 523

Text #1

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.1019. Min Act: -0.0001

Data Index: 299751 (C4 (Web Text))

Max Activating Token Index: 779

Text #2

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.1185. Min Act: -0.0001

Data Index: 1221294 (C4 (Web Text))

Max Activating Token Index: 243

Text #3

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.1080. Min Act: -0.0001

Data Index: 376283 (C4 (Web Text))

Max Activating Token Index: 926

Text #4

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.1027. Min Act: -0.0001

Data Index: 574029 (C4 (Web Text))

Max Activating Token Index: 316

Text #5

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.1086. Min Act: -0.0001

Data Index: 564977 (C4 (Web Text))

Max Activating Token Index: 264

Text #6

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.1024. Min Act: -0.0001

Data Index: 583520 (C4 (Web Text))

Max Activating Token Index: 352

Text #7

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.1025. Min Act: -0.0001

Data Index: 961938 (C4 (Web Text))

Max Activating Token Index: 428

Text #8

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0998. Min Act: -0.0001

Data Index: 112127 (C4 (Web Text))

Max Activating Token Index: 278

Text #9

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0933. Min Act: -0.0001

Data Index: 1645533 (Python Code)

Max Activating Token Index: 870

Text #10

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0893. Min Act: -0.0001

Data Index: 1355513 (C4 (Web Text))

Max Activating Token Index: 627

Text #11

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0916. Min Act: -0.0001

Data Index: 1273845 (C4 (Web Text))

Max Activating Token Index: 399

Text #12

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0928. Min Act: -0.0001

Data Index: 738087 (C4 (Web Text))

Max Activating Token Index: 116

Text #13

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0904. Min Act: -0.0001

Data Index: 1304509 (C4 (Web Text))

Max Activating Token Index: 58

Text #14

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0937. Min Act: -0.0001

Data Index: 1314693 (C4 (Web Text))

Max Activating Token Index: 93

Text #15

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0884. Min Act: -0.0001

Data Index: 1577286 (Python Code)

Max Activating Token Index: 769

Text #16

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0838. Min Act: -0.0001

Data Index: 253187 (C4 (Web Text))

Max Activating Token Index: 983

Text #17

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0800. Min Act: -0.0001

Data Index: 442231 (C4 (Web Text))

Max Activating Token Index: 812

Text #18

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0961. Min Act: -0.0001

Data Index: 63515 (C4 (Web Text))

Max Activating Token Index: 755

Text #19

Max Range: 0.1469. Min Range: -0.1469

Max Act: 0.0916. Min Act: -0.0001

Data Index: 646614 (C4 (Web Text))

Max Activating Token Index: 169