Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 2539 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1530. Min Act: -0.0002

Data Index: 1050018 (C4 (Web Text))

Max Activating Token Index: 103

Text #1

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1506. Min Act: -0.0002

Data Index: 1078759 (C4 (Web Text))

Max Activating Token Index: 83

Text #2

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1517. Min Act: -0.0002

Data Index: 955147 (C4 (Web Text))

Max Activating Token Index: 101

Text #3

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1361. Min Act: -0.0002

Data Index: 1183865 (C4 (Web Text))

Max Activating Token Index: 677

Text #4

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1304. Min Act: -0.0002

Data Index: 670489 (C4 (Web Text))

Max Activating Token Index: 56

Text #5

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1340. Min Act: -0.0002

Data Index: 367183 (C4 (Web Text))

Max Activating Token Index: 635

Text #6

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1278. Min Act: -0.0002

Data Index: 238104 (C4 (Web Text))

Max Activating Token Index: 782

Text #7

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1143. Min Act: -0.0002

Data Index: 665008 (C4 (Web Text))

Max Activating Token Index: 799

Text #8

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1172. Min Act: -0.0002

Data Index: 35866 (C4 (Web Text))

Max Activating Token Index: 922

Text #9

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1182. Min Act: -0.0001

Data Index: 1397183 (Python Code)

Max Activating Token Index: 579

Text #10

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1137. Min Act: -0.0002

Data Index: 340764 (C4 (Web Text))

Max Activating Token Index: 896

Text #11

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1202. Min Act: -0.0002

Data Index: 57718 (C4 (Web Text))

Max Activating Token Index: 554

Text #12

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1050. Min Act: -0.0001

Data Index: 1497190 (Python Code)

Max Activating Token Index: 632

Text #13

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1088. Min Act: -0.0002

Data Index: 84787 (C4 (Web Text))

Max Activating Token Index: 424

Text #14

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1127. Min Act: -0.0002

Data Index: 505438 (C4 (Web Text))

Max Activating Token Index: 224

Text #15

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1120. Min Act: -0.0002

Data Index: 817293 (C4 (Web Text))

Max Activating Token Index: 322

Text #16

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1063. Min Act: -0.0002

Data Index: 488005 (C4 (Web Text))

Max Activating Token Index: 217

Text #17

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1135. Min Act: -0.0002

Data Index: 1299677 (C4 (Web Text))

Max Activating Token Index: 957

Text #18

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1036. Min Act: -0.0002

Data Index: 274736 (C4 (Web Text))

Max Activating Token Index: 423

Text #19

Max Range: 0.1530. Min Range: -0.1530

Max Act: 0.1172. Min Act: -0.0002

Data Index: 1126724 (C4 (Web Text))

Max Activating Token Index: 373