Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 859 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0587. Min Act: -0.0003

Data Index: 888257 (C4 (Web Text))

Max Activating Token Index: 639

Text #1

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0530. Min Act: -0.0004

Data Index: 231512 (C4 (Web Text))

Max Activating Token Index: 516

Text #2

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0533. Min Act: -0.0004

Data Index: 853591 (C4 (Web Text))

Max Activating Token Index: 27

Text #3

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0528. Min Act: -0.0002

Data Index: 1655733 (Python Code)

Max Activating Token Index: 108

Text #4

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0519. Min Act: -0.0003

Data Index: 983134 (C4 (Web Text))

Max Activating Token Index: 124

Text #5

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0498. Min Act: -0.0003

Data Index: 1351351 (C4 (Web Text))

Max Activating Token Index: 329

Text #6

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0490. Min Act: -0.0003

Data Index: 636646 (C4 (Web Text))

Max Activating Token Index: 558

Text #7

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0471. Min Act: -0.0002

Data Index: 855503 (C4 (Web Text))

Max Activating Token Index: 175

Text #8

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0457. Min Act: -0.0003

Data Index: 928071 (C4 (Web Text))

Max Activating Token Index: 436

Text #9

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0459. Min Act: -0.0002

Data Index: 630350 (C4 (Web Text))

Max Activating Token Index: 26

Text #10

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0437. Min Act: -0.0003

Data Index: 673971 (C4 (Web Text))

Max Activating Token Index: 512

Text #11

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0451. Min Act: -0.0003

Data Index: 460715 (C4 (Web Text))

Max Activating Token Index: 430

Text #12

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0443. Min Act: -0.0003

Data Index: 632792 (C4 (Web Text))

Max Activating Token Index: 591

Text #13

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0450. Min Act: -0.0002

Data Index: 1037074 (C4 (Web Text))

Max Activating Token Index: 844

Text #14

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0445. Min Act: -0.0002

Data Index: 525618 (C4 (Web Text))

Max Activating Token Index: 302

Text #15

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0450. Min Act: -0.0003

Data Index: 568353 (C4 (Web Text))

Max Activating Token Index: 875

Text #16

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0439. Min Act: -0.0002

Data Index: 1349917 (C4 (Web Text))

Max Activating Token Index: 852

Text #17

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0462. Min Act: -0.0003

Data Index: 1618815 (Python Code)

Max Activating Token Index: 800

Text #18

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0433. Min Act: -0.0003

Data Index: 446227 (C4 (Web Text))

Max Activating Token Index: 746

Text #19

Max Range: 0.0587. Min Range: -0.0587

Max Act: 0.0441. Min Act: -0.0003

Data Index: 1070320 (C4 (Web Text))

Max Activating Token Index: 404