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

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

Neuron 2032 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.3973. Min Act: -0.0000

Data Index: 493781 (C4 (Web Text))

Max Activating Token Index: 1009

Text #1

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2647. Min Act: -0.0000

Data Index: 111590 (C4 (Web Text))

Max Activating Token Index: 939

Text #2

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2590. Min Act: -0.0000

Data Index: 684766 (C4 (Web Text))

Max Activating Token Index: 33

Text #3

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2063. Min Act: -0.0000

Data Index: 702616 (C4 (Web Text))

Max Activating Token Index: 874

Text #4

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2260. Min Act: -0.0000

Data Index: 658843 (C4 (Web Text))

Max Activating Token Index: 427

Text #5

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2277. Min Act: -0.0000

Data Index: 105203 (C4 (Web Text))

Max Activating Token Index: 591

Text #6

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2147. Min Act: -0.0000

Data Index: 164830 (C4 (Web Text))

Max Activating Token Index: 234

Text #7

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2103. Min Act: -0.0000

Data Index: 1525498 (Python Code)

Max Activating Token Index: 844

Text #8

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2201. Min Act: -0.0000

Data Index: 1043516 (C4 (Web Text))

Max Activating Token Index: 807

Text #9

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2183. Min Act: 0.0000

Data Index: 230243 (C4 (Web Text))

Max Activating Token Index: 370

Text #10

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.1932. Min Act: -0.0000

Data Index: 1399127 (Python Code)

Max Activating Token Index: 40

Text #11

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2050. Min Act: -0.0000

Data Index: 1490662 (Python Code)

Max Activating Token Index: 451

Text #12

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2142. Min Act: 0.0000

Data Index: 373099 (C4 (Web Text))

Max Activating Token Index: 455

Text #13

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2199. Min Act: -0.0000

Data Index: 1530115 (Python Code)

Max Activating Token Index: 629

Text #14

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2037. Min Act: -0.0000

Data Index: 1123954 (C4 (Web Text))

Max Activating Token Index: 484

Text #15

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.1920. Min Act: 0.0000

Data Index: 926728 (C4 (Web Text))

Max Activating Token Index: 281

Text #16

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.2203. Min Act: -0.0000

Data Index: 291309 (C4 (Web Text))

Max Activating Token Index: 818

Text #17

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.1947. Min Act: -0.0000

Data Index: 1407262 (Python Code)

Max Activating Token Index: 77

Text #18

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.1934. Min Act: -0.0000

Data Index: 1347740 (C4 (Web Text))

Max Activating Token Index: 446

Text #19

Max Range: 0.3973. Min Range: -0.3973

Max Act: 0.1945. Min Act: -0.0000

Data Index: 806595 (C4 (Web Text))

Max Activating Token Index: 656