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

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

Neuron 20 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.0560. Min Range: -1.0560

Max Act: 1.0560. Min Act: -0.0000

Data Index: 766006 (C4 (Web Text))

Max Activating Token Index: 55

Text #1

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.9019. Min Act: -0.0000

Data Index: 1030682 (C4 (Web Text))

Max Activating Token Index: 77

Text #2

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8936. Min Act: -0.0000

Data Index: 329497 (C4 (Web Text))

Max Activating Token Index: 37

Text #3

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8567. Min Act: -0.0000

Data Index: 561134 (C4 (Web Text))

Max Activating Token Index: 337

Text #4

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8697. Min Act: -0.0000

Data Index: 1146962 (C4 (Web Text))

Max Activating Token Index: 847

Text #5

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8170. Min Act: -0.0000

Data Index: 729779 (C4 (Web Text))

Max Activating Token Index: 900

Text #6

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8382. Min Act: -0.0000

Data Index: 1077514 (C4 (Web Text))

Max Activating Token Index: 281

Text #7

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8666. Min Act: -0.0000

Data Index: 646916 (C4 (Web Text))

Max Activating Token Index: 1004

Text #8

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8131. Min Act: -0.0000

Data Index: 470812 (C4 (Web Text))

Max Activating Token Index: 204

Text #9

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8605. Min Act: -0.0000

Data Index: 397129 (C4 (Web Text))

Max Activating Token Index: 851

Text #10

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8344. Min Act: -0.0000

Data Index: 14551 (C4 (Web Text))

Max Activating Token Index: 116

Text #11

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8511. Min Act: -0.0000

Data Index: 1171312 (C4 (Web Text))

Max Activating Token Index: 337

Text #12

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.7884. Min Act: -0.0000

Data Index: 1107949 (C4 (Web Text))

Max Activating Token Index: 581

Text #13

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.7932. Min Act: -0.0000

Data Index: 903067 (C4 (Web Text))

Max Activating Token Index: 1006

Text #14

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.8441. Min Act: -0.0000

Data Index: 1401127 (Python Code)

Max Activating Token Index: 279

Text #15

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.7477. Min Act: -0.0000

Data Index: 676874 (C4 (Web Text))

Max Activating Token Index: 844

Text #16

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.7472. Min Act: -0.0000

Data Index: 731123 (C4 (Web Text))

Max Activating Token Index: 990

Text #17

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.7661. Min Act: -0.0000

Data Index: 266663 (C4 (Web Text))

Max Activating Token Index: 732

Text #18

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.7703. Min Act: -0.0000

Data Index: 458011 (C4 (Web Text))

Max Activating Token Index: 279

Text #19

Max Range: 1.0560. Min Range: -1.0560

Max Act: 0.7604. Min Act: -0.0000

Data Index: 2826 (C4 (Web Text))

Max Activating Token Index: 168