Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 321 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2696. Min Act: -0.0002

Data Index: 956362 (C4 (Web Text))

Max Activating Token Index: 886

Text #1

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2355. Min Act: -0.0002

Data Index: 1406056 (Python Code)

Max Activating Token Index: 852

Text #2

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2196. Min Act: -0.0002

Data Index: 610108 (C4 (Web Text))

Max Activating Token Index: 176

Text #3

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2085. Min Act: -0.0002

Data Index: 654328 (C4 (Web Text))

Max Activating Token Index: 824

Text #4

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2119. Min Act: -0.0002

Data Index: 1586174 (Python Code)

Max Activating Token Index: 850

Text #5

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2055. Min Act: -0.0002

Data Index: 810403 (C4 (Web Text))

Max Activating Token Index: 865

Text #6

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2184. Min Act: -0.0002

Data Index: 502806 (C4 (Web Text))

Max Activating Token Index: 886

Text #7

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2055. Min Act: -0.0002

Data Index: 1037150 (C4 (Web Text))

Max Activating Token Index: 344

Text #8

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2095. Min Act: -0.0002

Data Index: 153008 (C4 (Web Text))

Max Activating Token Index: 684

Text #9

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1969. Min Act: -0.0002

Data Index: 1424578 (Python Code)

Max Activating Token Index: 986

Text #10

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.2133. Min Act: -0.0002

Data Index: 1410901 (Python Code)

Max Activating Token Index: 592

Text #11

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1947. Min Act: -0.0002

Data Index: 669656 (C4 (Web Text))

Max Activating Token Index: 1020

Text #12

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1815. Min Act: -0.0002

Data Index: 1042386 (C4 (Web Text))

Max Activating Token Index: 453

Text #13

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1946. Min Act: -0.0002

Data Index: 605820 (C4 (Web Text))

Max Activating Token Index: 375

Text #14

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1879. Min Act: -0.0002

Data Index: 947795 (C4 (Web Text))

Max Activating Token Index: 706

Text #15

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1879. Min Act: -0.0002

Data Index: 1154252 (C4 (Web Text))

Max Activating Token Index: 268

Text #16

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1911. Min Act: -0.0002

Data Index: 1129225 (C4 (Web Text))

Max Activating Token Index: 93

Text #17

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1978. Min Act: -0.0002

Data Index: 1441529 (Python Code)

Max Activating Token Index: 781

Text #18

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1944. Min Act: -0.0002

Data Index: 514466 (C4 (Web Text))

Max Activating Token Index: 414

Text #19

Max Range: 0.2696. Min Range: -0.2696

Max Act: 0.1822. Min Act: -0.0002

Data Index: 1466830 (Python Code)

Max Activating Token Index: 983