Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 900 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0782. Min Act: -0.0002

Data Index: 204049 (C4 (Web Text))

Max Activating Token Index: 875

Text #1

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0690. Min Act: -0.0002

Data Index: 170622 (C4 (Web Text))

Max Activating Token Index: 26

Text #2

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0670. Min Act: -0.0001

Data Index: 33803 (C4 (Web Text))

Max Activating Token Index: 105

Text #3

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0656. Min Act: -0.0002

Data Index: 406995 (C4 (Web Text))

Max Activating Token Index: 624

Text #4

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0598. Min Act: -0.0002

Data Index: 787453 (C4 (Web Text))

Max Activating Token Index: 986

Text #5

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0573. Min Act: -0.0002

Data Index: 140743 (C4 (Web Text))

Max Activating Token Index: 329

Text #6

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0587. Min Act: -0.0002

Data Index: 870649 (C4 (Web Text))

Max Activating Token Index: 693

Text #7

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0592. Min Act: -0.0002

Data Index: 1539103 (Python Code)

Max Activating Token Index: 196

Text #8

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0597. Min Act: -0.0002

Data Index: 600523 (C4 (Web Text))

Max Activating Token Index: 899

Text #9

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0586. Min Act: -0.0002

Data Index: 1200455 (C4 (Web Text))

Max Activating Token Index: 600

Text #10

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0583. Min Act: -0.0002

Data Index: 94021 (C4 (Web Text))

Max Activating Token Index: 87

Text #11

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0540. Min Act: -0.0002

Data Index: 990504 (C4 (Web Text))

Max Activating Token Index: 812

Text #12

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0585. Min Act: -0.0002

Data Index: 1634311 (Python Code)

Max Activating Token Index: 228

Text #13

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0560. Min Act: -0.0002

Data Index: 1260347 (C4 (Web Text))

Max Activating Token Index: 677

Text #14

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0545. Min Act: -0.0002

Data Index: 499348 (C4 (Web Text))

Max Activating Token Index: 982

Text #15

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0564. Min Act: -0.0002

Data Index: 1131510 (C4 (Web Text))

Max Activating Token Index: 297

Text #16

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0551. Min Act: -0.0002

Data Index: 357954 (C4 (Web Text))

Max Activating Token Index: 21

Text #17

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0546. Min Act: -0.0002

Data Index: 1544443 (Python Code)

Max Activating Token Index: 759

Text #18

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0541. Min Act: -0.0002

Data Index: 859619 (C4 (Web Text))

Max Activating Token Index: 901

Text #19

Max Range: 0.0782. Min Range: -0.0782

Max Act: 0.0550. Min Act: -0.0002

Data Index: 1591712 (Python Code)

Max Activating Token Index: 662