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

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

Neuron 3449 in Layer 6

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.9071. Min Act: -0.0000

Data Index: 1392403 (Python Code)

Max Activating Token Index: 919

Text #1

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.8178. Min Act: -0.0000

Data Index: 1517191 (Python Code)

Max Activating Token Index: 521

Text #2

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.8176. Min Act: -0.0000

Data Index: 1514414 (Python Code)

Max Activating Token Index: 28

Text #3

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.7206. Min Act: -0.0000

Data Index: 404941 (C4 (Web Text))

Max Activating Token Index: 773

Text #4

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.7482. Min Act: -0.0000

Data Index: 1427070 (Python Code)

Max Activating Token Index: 563

Text #5

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.7095. Min Act: -0.0000

Data Index: 395362 (C4 (Web Text))

Max Activating Token Index: 312

Text #6

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.7085. Min Act: -0.0000

Data Index: 227333 (C4 (Web Text))

Max Activating Token Index: 329

Text #7

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6888. Min Act: -0.0000

Data Index: 1226067 (C4 (Web Text))

Max Activating Token Index: 613

Text #8

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6792. Min Act: -0.0000

Data Index: 1246958 (C4 (Web Text))

Max Activating Token Index: 236

Text #9

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6900. Min Act: -0.0000

Data Index: 570767 (C4 (Web Text))

Max Activating Token Index: 780

Text #10

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.7162. Min Act: -0.0000

Data Index: 770387 (C4 (Web Text))

Max Activating Token Index: 934

Text #11

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6659. Min Act: -0.0000

Data Index: 1610951 (Python Code)

Max Activating Token Index: 668

Text #12

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6608. Min Act: -0.0000

Data Index: 858230 (C4 (Web Text))

Max Activating Token Index: 992

Text #13

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6877. Min Act: -0.0000

Data Index: 1479998 (Python Code)

Max Activating Token Index: 826

Text #14

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6739. Min Act: -0.0000

Data Index: 622719 (C4 (Web Text))

Max Activating Token Index: 563

Text #15

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6440. Min Act: -0.0000

Data Index: 927805 (C4 (Web Text))

Max Activating Token Index: 346

Text #16

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6647. Min Act: -0.0000

Data Index: 1372054 (Python Code)

Max Activating Token Index: 644

Text #17

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6413. Min Act: -0.0000

Data Index: 628252 (C4 (Web Text))

Max Activating Token Index: 303

Text #18

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6151. Min Act: -0.0000

Data Index: 637270 (C4 (Web Text))

Max Activating Token Index: 650

Text #19

Max Range: 0.9071. Min Range: -0.9071

Max Act: 0.6584. Min Act: -0.0000

Data Index: 1205392 (C4 (Web Text))

Max Activating Token Index: 702