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

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

Neuron 2032 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0095. Min Act: -0.0005

Data Index: 320566 (C4 (Web Text))

Max Activating Token Index: 147

Text #1

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0090. Min Act: -0.0006

Data Index: 1558651 (Python Code)

Max Activating Token Index: 133

Text #2

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0089. Min Act: -0.0006

Data Index: 1636245 (Python Code)

Max Activating Token Index: 342

Text #3

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0086. Min Act: -0.0006

Data Index: 553711 (C4 (Web Text))

Max Activating Token Index: 1023

Text #4

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0085. Min Act: -0.0005

Data Index: 1226472 (C4 (Web Text))

Max Activating Token Index: 1023

Text #5

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0084. Min Act: -0.0005

Data Index: 668400 (C4 (Web Text))

Max Activating Token Index: 872

Text #6

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0084. Min Act: -0.0005

Data Index: 1090822 (C4 (Web Text))

Max Activating Token Index: 391

Text #7

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0083. Min Act: -0.0005

Data Index: 1408154 (Python Code)

Max Activating Token Index: 17

Text #8

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0083. Min Act: -0.0006

Data Index: 590655 (C4 (Web Text))

Max Activating Token Index: 107

Text #9

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0083. Min Act: -0.0006

Data Index: 689602 (C4 (Web Text))

Max Activating Token Index: 1023

Text #10

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0081. Min Act: -0.0005

Data Index: 925772 (C4 (Web Text))

Max Activating Token Index: 907

Text #11

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0079. Min Act: -0.0006

Data Index: 471457 (C4 (Web Text))

Max Activating Token Index: 213

Text #12

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0081. Min Act: -0.0005

Data Index: 401740 (C4 (Web Text))

Max Activating Token Index: 1023

Text #13

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0079. Min Act: -0.0006

Data Index: 232044 (C4 (Web Text))

Max Activating Token Index: 870

Text #14

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0078. Min Act: -0.0006

Data Index: 339193 (C4 (Web Text))

Max Activating Token Index: 327

Text #15

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0077. Min Act: -0.0005

Data Index: 85006 (C4 (Web Text))

Max Activating Token Index: 386

Text #16

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0073. Min Act: -0.0006

Data Index: 1570866 (Python Code)

Max Activating Token Index: 190

Text #17

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0075. Min Act: -0.0006

Data Index: 1427661 (Python Code)

Max Activating Token Index: 146

Text #18

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0074. Min Act: -0.0006

Data Index: 1220520 (C4 (Web Text))

Max Activating Token Index: 1023

Text #19

Max Range: 0.0095. Min Range: -0.0095

Max Act: 0.0073. Min Act: -0.0006

Data Index: 1422237 (Python Code)

Max Activating Token Index: 50