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

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

Neuron 3585 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.1012. Min Act: -0.0001

Data Index: 1138457 (C4 (Web Text))

Max Activating Token Index: 334

Text #1

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.1027. Min Act: -0.0001

Data Index: 805468 (C4 (Web Text))

Max Activating Token Index: 449

Text #2

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0922. Min Act: -0.0001

Data Index: 1027309 (C4 (Web Text))

Max Activating Token Index: 884

Text #3

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0941. Min Act: -0.0001

Data Index: 1595766 (Python Code)

Max Activating Token Index: 77

Text #4

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0913. Min Act: -0.0001

Data Index: 1481988 (Python Code)

Max Activating Token Index: 408

Text #5

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0854. Min Act: -0.0001

Data Index: 1132091 (C4 (Web Text))

Max Activating Token Index: 159

Text #6

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0884. Min Act: -0.0001

Data Index: 720717 (C4 (Web Text))

Max Activating Token Index: 866

Text #7

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0871. Min Act: -0.0001

Data Index: 1377898 (Python Code)

Max Activating Token Index: 491

Text #8

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0869. Min Act: -0.0001

Data Index: 494435 (C4 (Web Text))

Max Activating Token Index: 377

Text #9

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0870. Min Act: -0.0000

Data Index: 47326 (C4 (Web Text))

Max Activating Token Index: 353

Text #10

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0823. Min Act: -0.0001

Data Index: 1224623 (C4 (Web Text))

Max Activating Token Index: 58

Text #11

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0815. Min Act: -0.0001

Data Index: 341520 (C4 (Web Text))

Max Activating Token Index: 1007

Text #12

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0804. Min Act: -0.0001

Data Index: 1475395 (Python Code)

Max Activating Token Index: 789

Text #13

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0843. Min Act: -0.0001

Data Index: 186239 (C4 (Web Text))

Max Activating Token Index: 288

Text #14

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0779. Min Act: -0.0001

Data Index: 1411764 (Python Code)

Max Activating Token Index: 721

Text #15

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0803. Min Act: -0.0001

Data Index: 1642573 (Python Code)

Max Activating Token Index: 1011

Text #16

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0802. Min Act: -0.0001

Data Index: 408589 (C4 (Web Text))

Max Activating Token Index: 341

Text #17

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0785. Min Act: -0.0001

Data Index: 196294 (C4 (Web Text))

Max Activating Token Index: 437

Text #18

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0758. Min Act: -0.0001

Data Index: 1451077 (Python Code)

Max Activating Token Index: 102

Text #19

Max Range: 0.1027. Min Range: -0.1027

Max Act: 0.0795. Min Act: -0.0001

Data Index: 432030 (C4 (Web Text))

Max Activating Token Index: 71