Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 3046 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.6316. Min Act: -0.0003

Data Index: 294515 (C4 (Web Text))

Max Activating Token Index: 1007

Text #1

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.6038. Min Act: -0.0003

Data Index: 1514085 (Python Code)

Max Activating Token Index: 985

Text #2

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5758. Min Act: -0.0003

Data Index: 352744 (C4 (Web Text))

Max Activating Token Index: 63

Text #3

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5592. Min Act: -0.0003

Data Index: 1236125 (C4 (Web Text))

Max Activating Token Index: 106

Text #4

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5781. Min Act: -0.0003

Data Index: 155852 (C4 (Web Text))

Max Activating Token Index: 853

Text #5

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5734. Min Act: -0.0003

Data Index: 104974 (C4 (Web Text))

Max Activating Token Index: 1018

Text #6

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5513. Min Act: -0.0003

Data Index: 1264858 (C4 (Web Text))

Max Activating Token Index: 37

Text #7

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5067. Min Act: -0.0003

Data Index: 1645134 (Python Code)

Max Activating Token Index: 1008

Text #8

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5333. Min Act: -0.0003

Data Index: 1308698 (C4 (Web Text))

Max Activating Token Index: 925

Text #9

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5538. Min Act: -0.0003

Data Index: 1150107 (C4 (Web Text))

Max Activating Token Index: 917

Text #10

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.4997. Min Act: -0.0003

Data Index: 172695 (C4 (Web Text))

Max Activating Token Index: 771

Text #11

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5437. Min Act: -0.0003

Data Index: 1042007 (C4 (Web Text))

Max Activating Token Index: 120

Text #12

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5105. Min Act: -0.0003

Data Index: 810516 (C4 (Web Text))

Max Activating Token Index: 360

Text #13

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5522. Min Act: -0.0003

Data Index: 399150 (C4 (Web Text))

Max Activating Token Index: 172

Text #14

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5333. Min Act: -0.0003

Data Index: 741426 (C4 (Web Text))

Max Activating Token Index: 889

Text #15

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.4969. Min Act: -0.0003

Data Index: 6475 (C4 (Web Text))

Max Activating Token Index: 471

Text #16

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5141. Min Act: -0.0003

Data Index: 1113580 (C4 (Web Text))

Max Activating Token Index: 104

Text #17

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5117. Min Act: -0.0003

Data Index: 17409 (C4 (Web Text))

Max Activating Token Index: 676

Text #18

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.5131. Min Act: -0.0003

Data Index: 1549151 (Python Code)

Max Activating Token Index: 304

Text #19

Max Range: 0.6316. Min Range: -0.6316

Max Act: 0.4618. Min Act: -0.0003

Data Index: 636123 (C4 (Web Text))

Max Activating Token Index: 394