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

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

Neuron 1591 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: 0.8161. Min Range: -0.8161

Max Act: 0.8161. Min Act: -0.0001

Data Index: 714486 (C4 (Web Text))

Max Activating Token Index: 571

Text #1

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.7494. Min Act: -0.0001

Data Index: 849313 (C4 (Web Text))

Max Activating Token Index: 61

Text #2

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.7146. Min Act: -0.0001

Data Index: 95263 (C4 (Web Text))

Max Activating Token Index: 997

Text #3

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.7057. Min Act: -0.0001

Data Index: 88869 (C4 (Web Text))

Max Activating Token Index: 936

Text #4

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.7346. Min Act: -0.0001

Data Index: 940399 (C4 (Web Text))

Max Activating Token Index: 172

Text #5

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6647. Min Act: -0.0001

Data Index: 289997 (C4 (Web Text))

Max Activating Token Index: 940

Text #6

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6780. Min Act: -0.0001

Data Index: 141091 (C4 (Web Text))

Max Activating Token Index: 8

Text #7

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6679. Min Act: -0.0001

Data Index: 290039 (C4 (Web Text))

Max Activating Token Index: 679

Text #8

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6553. Min Act: -0.0001

Data Index: 669639 (C4 (Web Text))

Max Activating Token Index: 811

Text #9

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6496. Min Act: -0.0001

Data Index: 107196 (C4 (Web Text))

Max Activating Token Index: 90

Text #10

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6572. Min Act: -0.0001

Data Index: 1059456 (C4 (Web Text))

Max Activating Token Index: 328

Text #11

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.7040. Min Act: -0.0001

Data Index: 604484 (C4 (Web Text))

Max Activating Token Index: 880

Text #12

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6153. Min Act: -0.0001

Data Index: 104470 (C4 (Web Text))

Max Activating Token Index: 593

Text #13

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6702. Min Act: -0.0001

Data Index: 1376645 (Python Code)

Max Activating Token Index: 508

Text #14

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6746. Min Act: -0.0001

Data Index: 1057080 (C4 (Web Text))

Max Activating Token Index: 106

Text #15

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6427. Min Act: -0.0001

Data Index: 565086 (C4 (Web Text))

Max Activating Token Index: 119

Text #16

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6472. Min Act: -0.0001

Data Index: 26886 (C4 (Web Text))

Max Activating Token Index: 872

Text #17

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6232. Min Act: -0.0001

Data Index: 1391524 (Python Code)

Max Activating Token Index: 627

Text #18

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6415. Min Act: -0.0001

Data Index: 1157592 (C4 (Web Text))

Max Activating Token Index: 676

Text #19

Max Range: 0.8161. Min Range: -0.8161

Max Act: 0.6432. Min Act: -0.0001

Data Index: 1519936 (Python Code)

Max Activating Token Index: 113