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

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

Neuron 2707 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: 4.1589. Min Range: -4.1589

Max Act: 4.1589. Min Act: -0.0000

Data Index: 966912 (C4 (Web Text))

Max Activating Token Index: 469

Text #1

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.7566. Min Act: -0.0000

Data Index: 1482329 (Python Code)

Max Activating Token Index: 524

Text #2

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.8855. Min Act: -0.0000

Data Index: 416885 (C4 (Web Text))

Max Activating Token Index: 420

Text #3

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.9218. Min Act: -0.0000

Data Index: 131820 (C4 (Web Text))

Max Activating Token Index: 113

Text #4

Max Range: 4.1589. Min Range: -4.1589

Max Act: 4.0284. Min Act: -0.0000

Data Index: 663115 (C4 (Web Text))

Max Activating Token Index: 381

Text #5

Max Range: 4.1589. Min Range: -4.1589

Max Act: 2.9911. Min Act: -0.0000

Data Index: 1608795 (Python Code)

Max Activating Token Index: 681

Text #6

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.2752. Min Act: -0.0000

Data Index: 1421010 (Python Code)

Max Activating Token Index: 746

Text #7

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.5078. Min Act: -0.0000

Data Index: 387828 (C4 (Web Text))

Max Activating Token Index: 240

Text #8

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.3267. Min Act: -0.0000

Data Index: 1592963 (Python Code)

Max Activating Token Index: 531

Text #9

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.4672. Min Act: -0.0000

Data Index: 778546 (C4 (Web Text))

Max Activating Token Index: 398

Text #10

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.3994. Min Act: -0.0000

Data Index: 1268921 (C4 (Web Text))

Max Activating Token Index: 595

Text #11

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.1896. Min Act: -0.0000

Data Index: 1476922 (Python Code)

Max Activating Token Index: 440

Text #12

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.1597. Min Act: -0.0000

Data Index: 1324653 (C4 (Web Text))

Max Activating Token Index: 364

Text #13

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.1061. Min Act: -0.0000

Data Index: 1142385 (C4 (Web Text))

Max Activating Token Index: 410

Text #14

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.3906. Min Act: -0.0000

Data Index: 458259 (C4 (Web Text))

Max Activating Token Index: 161

Text #15

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.1436. Min Act: -0.0000

Data Index: 1295610 (C4 (Web Text))

Max Activating Token Index: 450

Text #16

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.3047. Min Act: -0.0000

Data Index: 477487 (C4 (Web Text))

Max Activating Token Index: 955

Text #17

Max Range: 4.1589. Min Range: -4.1589

Max Act: 2.9795. Min Act: -0.0000

Data Index: 876888 (C4 (Web Text))

Max Activating Token Index: 357

Text #18

Max Range: 4.1589. Min Range: -4.1589

Max Act: 2.9998. Min Act: -0.0000

Data Index: 905719 (C4 (Web Text))

Max Activating Token Index: 748

Text #19

Max Range: 4.1589. Min Range: -4.1589

Max Act: 3.0978. Min Act: -0.0000

Data Index: 1611765 (Python Code)

Max Activating Token Index: 941