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

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

Neuron 4714 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.9834. Min Act: -0.0001

Data Index: 1456378 (Python Code)

Max Activating Token Index: 460

Text #1

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8478. Min Act: -0.0001

Data Index: 238307 (C4 (Web Text))

Max Activating Token Index: 653

Text #2

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8789. Min Act: -0.0001

Data Index: 938177 (C4 (Web Text))

Max Activating Token Index: 168

Text #3

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.9141. Min Act: -0.0001

Data Index: 1394443 (Python Code)

Max Activating Token Index: 3

Text #4

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8332. Min Act: -0.0001

Data Index: 945721 (C4 (Web Text))

Max Activating Token Index: 633

Text #5

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8344. Min Act: -0.0001

Data Index: 1311450 (C4 (Web Text))

Max Activating Token Index: 881

Text #6

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8073. Min Act: -0.0001

Data Index: 1468892 (Python Code)

Max Activating Token Index: 755

Text #7

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8672. Min Act: -0.0001

Data Index: 996914 (C4 (Web Text))

Max Activating Token Index: 475

Text #8

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.7621. Min Act: -0.0001

Data Index: 1033635 (C4 (Web Text))

Max Activating Token Index: 472

Text #9

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8575. Min Act: -0.0001

Data Index: 130896 (C4 (Web Text))

Max Activating Token Index: 939

Text #10

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.7989. Min Act: -0.0001

Data Index: 153015 (C4 (Web Text))

Max Activating Token Index: 808

Text #11

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.7866. Min Act: -0.0000

Data Index: 1620196 (Python Code)

Max Activating Token Index: 86

Text #12

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.7930. Min Act: -0.0001

Data Index: 619104 (C4 (Web Text))

Max Activating Token Index: 921

Text #13

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.7322. Min Act: -0.0001

Data Index: 504629 (C4 (Web Text))

Max Activating Token Index: 836

Text #14

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8206. Min Act: -0.0001

Data Index: 1549766 (Python Code)

Max Activating Token Index: 915

Text #15

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8242. Min Act: -0.0000

Data Index: 1467200 (Python Code)

Max Activating Token Index: 943

Text #16

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.8420. Min Act: -0.0001

Data Index: 1509135 (Python Code)

Max Activating Token Index: 157

Text #17

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.6871. Min Act: -0.0001

Data Index: 377250 (C4 (Web Text))

Max Activating Token Index: 155

Text #18

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.7603. Min Act: -0.0001

Data Index: 882042 (C4 (Web Text))

Max Activating Token Index: 16

Text #19

Max Range: 0.9834. Min Range: -0.9834

Max Act: 0.7603. Min Act: -0.0001

Data Index: 134469 (C4 (Web Text))

Max Activating Token Index: 537