Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 2971 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1405. Min Act: -0.0001

Data Index: 631836 (C4 (Web Text))

Max Activating Token Index: 925

Text #1

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1263. Min Act: -0.0001

Data Index: 315000 (C4 (Web Text))

Max Activating Token Index: 51

Text #2

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1261. Min Act: -0.0001

Data Index: 320912 (C4 (Web Text))

Max Activating Token Index: 66

Text #3

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1248. Min Act: -0.0001

Data Index: 1130934 (C4 (Web Text))

Max Activating Token Index: 165

Text #4

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1196. Min Act: -0.0001

Data Index: 1190439 (C4 (Web Text))

Max Activating Token Index: 62

Text #5

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1171. Min Act: -0.0001

Data Index: 1416059 (Python Code)

Max Activating Token Index: 742

Text #6

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1252. Min Act: -0.0001

Data Index: 14537 (C4 (Web Text))

Max Activating Token Index: 368

Text #7

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1141. Min Act: -0.0001

Data Index: 509002 (C4 (Web Text))

Max Activating Token Index: 140

Text #8

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1143. Min Act: -0.0001

Data Index: 1004680 (C4 (Web Text))

Max Activating Token Index: 703

Text #9

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1135. Min Act: -0.0001

Data Index: 575234 (C4 (Web Text))

Max Activating Token Index: 212

Text #10

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1129. Min Act: -0.0001

Data Index: 1108509 (C4 (Web Text))

Max Activating Token Index: 140

Text #11

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1119. Min Act: -0.0001

Data Index: 1404378 (Python Code)

Max Activating Token Index: 461

Text #12

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1138. Min Act: -0.0001

Data Index: 493030 (C4 (Web Text))

Max Activating Token Index: 166

Text #13

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1082. Min Act: -0.0001

Data Index: 1522117 (Python Code)

Max Activating Token Index: 667

Text #14

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1083. Min Act: -0.0001

Data Index: 281470 (C4 (Web Text))

Max Activating Token Index: 808

Text #15

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1081. Min Act: -0.0001

Data Index: 1477283 (Python Code)

Max Activating Token Index: 458

Text #16

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1082. Min Act: -0.0001

Data Index: 51695 (C4 (Web Text))

Max Activating Token Index: 192

Text #17

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1072. Min Act: -0.0001

Data Index: 288528 (C4 (Web Text))

Max Activating Token Index: 678

Text #18

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1064. Min Act: -0.0001

Data Index: 100979 (C4 (Web Text))

Max Activating Token Index: 967

Text #19

Max Range: 0.1405. Min Range: -0.1405

Max Act: 0.1069. Min Act: -0.0001

Data Index: 426815 (C4 (Web Text))

Max Activating Token Index: 122