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

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

Neuron 305 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: 0.0754. Min Range: -0.0754

Max Act: 0.0754. Min Act: -0.0002

Data Index: 1136659 (C4 (Web Text))

Max Activating Token Index: 1010

Text #1

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0709. Min Act: -0.0002

Data Index: 312506 (C4 (Web Text))

Max Activating Token Index: 979

Text #2

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0702. Min Act: -0.0002

Data Index: 712395 (C4 (Web Text))

Max Activating Token Index: 459

Text #3

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0695. Min Act: -0.0002

Data Index: 192360 (C4 (Web Text))

Max Activating Token Index: 248

Text #4

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0660. Min Act: -0.0002

Data Index: 563964 (C4 (Web Text))

Max Activating Token Index: 349

Text #5

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0667. Min Act: -0.0002

Data Index: 1569397 (Python Code)

Max Activating Token Index: 65

Text #6

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0682. Min Act: -0.0002

Data Index: 174400 (C4 (Web Text))

Max Activating Token Index: 1001

Text #7

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0669. Min Act: -0.0002

Data Index: 50567 (C4 (Web Text))

Max Activating Token Index: 363

Text #8

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0626. Min Act: -0.0002

Data Index: 1011424 (C4 (Web Text))

Max Activating Token Index: 509

Text #9

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0655. Min Act: -0.0002

Data Index: 74157 (C4 (Web Text))

Max Activating Token Index: 928

Text #10

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0631. Min Act: -0.0002

Data Index: 956502 (C4 (Web Text))

Max Activating Token Index: 493

Text #11

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0674. Min Act: -0.0002

Data Index: 264397 (C4 (Web Text))

Max Activating Token Index: 288

Text #12

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0618. Min Act: -0.0002

Data Index: 1637461 (Python Code)

Max Activating Token Index: 854

Text #13

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0640. Min Act: -0.0002

Data Index: 717875 (C4 (Web Text))

Max Activating Token Index: 958

Text #14

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0632. Min Act: -0.0002

Data Index: 1555576 (Python Code)

Max Activating Token Index: 346

Text #15

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0624. Min Act: -0.0002

Data Index: 742121 (C4 (Web Text))

Max Activating Token Index: 1009

Text #16

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0631. Min Act: -0.0002

Data Index: 1487601 (Python Code)

Max Activating Token Index: 74

Text #17

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0600. Min Act: -0.0002

Data Index: 850066 (C4 (Web Text))

Max Activating Token Index: 687

Text #18

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0617. Min Act: -0.0002

Data Index: 635308 (C4 (Web Text))

Max Activating Token Index: 195

Text #19

Max Range: 0.0754. Min Range: -0.0754

Max Act: 0.0601. Min Act: -0.0002

Data Index: 214425 (C4 (Web Text))

Max Activating Token Index: 902