Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 5650 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 5.2009. Min Range: -5.2009

Max Act: 5.2009. Min Act: -0.0000

Data Index: 1075247 (C4 (Web Text))

Max Activating Token Index: 910

Text #1

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.9096. Min Act: -0.0000

Data Index: 1155173 (C4 (Web Text))

Max Activating Token Index: 925

Text #2

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.7671. Min Act: -0.0000

Data Index: 1575044 (Python Code)

Max Activating Token Index: 3

Text #3

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.6566. Min Act: -0.0000

Data Index: 1634365 (Python Code)

Max Activating Token Index: 299

Text #4

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.3601. Min Act: -0.0000

Data Index: 793422 (C4 (Web Text))

Max Activating Token Index: 832

Text #5

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.4977. Min Act: -0.0000

Data Index: 159904 (C4 (Web Text))

Max Activating Token Index: 114

Text #6

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.4210. Min Act: -0.0000

Data Index: 267254 (C4 (Web Text))

Max Activating Token Index: 126

Text #7

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.3551. Min Act: -0.0000

Data Index: 1191496 (C4 (Web Text))

Max Activating Token Index: 93

Text #8

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.3227. Min Act: -0.0000

Data Index: 706188 (C4 (Web Text))

Max Activating Token Index: 529

Text #9

Max Range: 5.2009. Min Range: -5.2009

Max Act: 3.9632. Min Act: -0.0000

Data Index: 1472161 (Python Code)

Max Activating Token Index: 984

Text #10

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.2223. Min Act: -0.0000

Data Index: 14326 (C4 (Web Text))

Max Activating Token Index: 492

Text #11

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.2025. Min Act: 0.0000

Data Index: 616400 (C4 (Web Text))

Max Activating Token Index: 739

Text #12

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.1725. Min Act: -0.0000

Data Index: 1430834 (Python Code)

Max Activating Token Index: 856

Text #13

Max Range: 5.2009. Min Range: -5.2009

Max Act: 3.9774. Min Act: -0.0000

Data Index: 1027923 (C4 (Web Text))

Max Activating Token Index: 294

Text #14

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.0805. Min Act: -0.0000

Data Index: 1123028 (C4 (Web Text))

Max Activating Token Index: 171

Text #15

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.1960. Min Act: -0.0000

Data Index: 1518951 (Python Code)

Max Activating Token Index: 689

Text #16

Max Range: 5.2009. Min Range: -5.2009

Max Act: 4.0420. Min Act: -0.0000

Data Index: 197213 (C4 (Web Text))

Max Activating Token Index: 247

Text #17

Max Range: 5.2009. Min Range: -5.2009

Max Act: 3.8333. Min Act: -0.0000

Data Index: 1591849 (Python Code)

Max Activating Token Index: 839

Text #18

Max Range: 5.2009. Min Range: -5.2009

Max Act: 3.9642. Min Act: -0.0000

Data Index: 1168158 (C4 (Web Text))

Max Activating Token Index: 379

Text #19

Max Range: 5.2009. Min Range: -5.2009

Max Act: 3.9232. Min Act: -0.0000

Data Index: 493325 (C4 (Web Text))

Max Activating Token Index: 817