Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 4556 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.6721. Min Act: -0.0000

Data Index: 1403975 (Python Code)

Max Activating Token Index: 213

Text #1

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.6637. Min Act: -0.0000

Data Index: 410132 (C4 (Web Text))

Max Activating Token Index: 471

Text #2

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.6691. Min Act: -0.0000

Data Index: 1592978 (Python Code)

Max Activating Token Index: 913

Text #3

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.6506. Min Act: -0.0000

Data Index: 1324550 (C4 (Web Text))

Max Activating Token Index: 319

Text #4

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.6474. Min Act: -0.0000

Data Index: 1096961 (C4 (Web Text))

Max Activating Token Index: 20

Text #5

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.6142. Min Act: -0.0000

Data Index: 585136 (C4 (Web Text))

Max Activating Token Index: 120

Text #6

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5791. Min Act: -0.0000

Data Index: 383668 (C4 (Web Text))

Max Activating Token Index: 202

Text #7

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5861. Min Act: -0.0000

Data Index: 445746 (C4 (Web Text))

Max Activating Token Index: 71

Text #8

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.6217. Min Act: -0.0000

Data Index: 660573 (C4 (Web Text))

Max Activating Token Index: 950

Text #9

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5442. Min Act: -0.0000

Data Index: 1649192 (Python Code)

Max Activating Token Index: 695

Text #10

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5805. Min Act: -0.0000

Data Index: 122876 (C4 (Web Text))

Max Activating Token Index: 128

Text #11

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5795. Min Act: -0.0000

Data Index: 1419365 (Python Code)

Max Activating Token Index: 623

Text #12

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5582. Min Act: -0.0000

Data Index: 542425 (C4 (Web Text))

Max Activating Token Index: 727

Text #13

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.6176. Min Act: -0.0000

Data Index: 1186941 (C4 (Web Text))

Max Activating Token Index: 761

Text #14

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5943. Min Act: -0.0000

Data Index: 16536 (C4 (Web Text))

Max Activating Token Index: 554

Text #15

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5585. Min Act: -0.0000

Data Index: 66063 (C4 (Web Text))

Max Activating Token Index: 91

Text #16

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5311. Min Act: -0.0000

Data Index: 661640 (C4 (Web Text))

Max Activating Token Index: 890

Text #17

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5259. Min Act: -0.0000

Data Index: 1485085 (Python Code)

Max Activating Token Index: 832

Text #18

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.5219. Min Act: -0.0000

Data Index: 1145271 (C4 (Web Text))

Max Activating Token Index: 764

Text #19

Max Range: 0.6721. Min Range: -0.6721

Max Act: 0.4950. Min Act: -0.0000

Data Index: 885290 (C4 (Web Text))

Max Activating Token Index: 404