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

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

Neuron 2628 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.1087. Min Range: -1.1087

Max Act: 1.0160. Min Act: -0.0001

Data Index: 1250491 (C4 (Web Text))

Max Activating Token Index: 811

Text #1

Max Range: 1.1087. Min Range: -1.1087

Max Act: 1.1087. Min Act: -0.0001

Data Index: 1036420 (C4 (Web Text))

Max Activating Token Index: 365

Text #2

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.9000. Min Act: -0.0001

Data Index: 1608250 (Python Code)

Max Activating Token Index: 905

Text #3

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.9337. Min Act: -0.0001

Data Index: 1384359 (Python Code)

Max Activating Token Index: 291

Text #4

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.9722. Min Act: -0.0001

Data Index: 937831 (C4 (Web Text))

Max Activating Token Index: 656

Text #5

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.9941. Min Act: -0.0001

Data Index: 1208047 (C4 (Web Text))

Max Activating Token Index: 893

Text #6

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.9157. Min Act: -0.0001

Data Index: 1129745 (C4 (Web Text))

Max Activating Token Index: 810

Text #7

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.8884. Min Act: -0.0001

Data Index: 341821 (C4 (Web Text))

Max Activating Token Index: 18

Text #8

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.9460. Min Act: -0.0001

Data Index: 1282317 (C4 (Web Text))

Max Activating Token Index: 36

Text #9

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.8634. Min Act: -0.0001

Data Index: 1330588 (C4 (Web Text))

Max Activating Token Index: 585

Text #10

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.8185. Min Act: -0.0001

Data Index: 898816 (C4 (Web Text))

Max Activating Token Index: 418

Text #11

Max Range: 1.1087. Min Range: -1.1087

Max Act: 1.0958. Min Act: -0.0001

Data Index: 517491 (C4 (Web Text))

Max Activating Token Index: 101

Text #12

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.6902. Min Act: -0.0001

Data Index: 1410149 (Python Code)

Max Activating Token Index: 767

Text #13

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.6846. Min Act: -0.0001

Data Index: 1469291 (Python Code)

Max Activating Token Index: 225

Text #14

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.7105. Min Act: -0.0001

Data Index: 907206 (C4 (Web Text))

Max Activating Token Index: 196

Text #15

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.6588. Min Act: -0.0001

Data Index: 530744 (C4 (Web Text))

Max Activating Token Index: 202

Text #16

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.7529. Min Act: -0.0001

Data Index: 540671 (C4 (Web Text))

Max Activating Token Index: 348

Text #17

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.8021. Min Act: -0.0001

Data Index: 941203 (C4 (Web Text))

Max Activating Token Index: 918

Text #18

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.7418. Min Act: -0.0001

Data Index: 911004 (C4 (Web Text))

Max Activating Token Index: 769

Text #19

Max Range: 1.1087. Min Range: -1.1087

Max Act: 0.7203. Min Act: -0.0001

Data Index: 536414 (C4 (Web Text))

Max Activating Token Index: 243