Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 797 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1138. Min Act: -0.0004

Data Index: 1048843 (C4 (Web Text))

Max Activating Token Index: 251

Text #1

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1121. Min Act: -0.0003

Data Index: 553376 (C4 (Web Text))

Max Activating Token Index: 435

Text #2

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1089. Min Act: -0.0004

Data Index: 509253 (C4 (Web Text))

Max Activating Token Index: 362

Text #3

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1094. Min Act: -0.0003

Data Index: 32833 (C4 (Web Text))

Max Activating Token Index: 87

Text #4

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1104. Min Act: -0.0003

Data Index: 1045636 (C4 (Web Text))

Max Activating Token Index: 40

Text #5

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1021. Min Act: -0.0004

Data Index: 306523 (C4 (Web Text))

Max Activating Token Index: 860

Text #6

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1094. Min Act: -0.0003

Data Index: 376395 (C4 (Web Text))

Max Activating Token Index: 635

Text #7

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1098. Min Act: -0.0004

Data Index: 820915 (C4 (Web Text))

Max Activating Token Index: 582

Text #8

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1049. Min Act: -0.0003

Data Index: 1200847 (C4 (Web Text))

Max Activating Token Index: 644

Text #9

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1083. Min Act: -0.0003

Data Index: 958464 (C4 (Web Text))

Max Activating Token Index: 417

Text #10

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1061. Min Act: -0.0004

Data Index: 1608395 (Python Code)

Max Activating Token Index: 923

Text #11

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1036. Min Act: -0.0004

Data Index: 669921 (C4 (Web Text))

Max Activating Token Index: 541

Text #12

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1037. Min Act: -0.0004

Data Index: 130952 (C4 (Web Text))

Max Activating Token Index: 950

Text #13

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1046. Min Act: -0.0004

Data Index: 553628 (C4 (Web Text))

Max Activating Token Index: 194

Text #14

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1043. Min Act: -0.0003

Data Index: 450628 (C4 (Web Text))

Max Activating Token Index: 196

Text #15

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.0968. Min Act: -0.0003

Data Index: 1541732 (Python Code)

Max Activating Token Index: 327

Text #16

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.1001. Min Act: -0.0003

Data Index: 679895 (C4 (Web Text))

Max Activating Token Index: 441

Text #17

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.0991. Min Act: -0.0004

Data Index: 87266 (C4 (Web Text))

Max Activating Token Index: 717

Text #18

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.0981. Min Act: -0.0004

Data Index: 362708 (C4 (Web Text))

Max Activating Token Index: 604

Text #19

Max Range: 0.1138. Min Range: -0.1138

Max Act: 0.0940. Min Act: -0.0004

Data Index: 716035 (C4 (Web Text))

Max Activating Token Index: 654