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

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

Neuron 4522 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.0522. Min Range: -1.0522

Max Act: 1.0522. Min Act: -0.0000

Data Index: 337177 (C4 (Web Text))

Max Activating Token Index: 904

Text #1

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.8612. Min Act: -0.0000

Data Index: 1528055 (Python Code)

Max Activating Token Index: 84

Text #2

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.8308. Min Act: -0.0000

Data Index: 483010 (C4 (Web Text))

Max Activating Token Index: 76

Text #3

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.7321. Min Act: -0.0000

Data Index: 1303934 (C4 (Web Text))

Max Activating Token Index: 282

Text #4

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.7638. Min Act: -0.0000

Data Index: 515249 (C4 (Web Text))

Max Activating Token Index: 828

Text #5

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.7667. Min Act: -0.0000

Data Index: 1600299 (Python Code)

Max Activating Token Index: 383

Text #6

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.7605. Min Act: -0.0000

Data Index: 785973 (C4 (Web Text))

Max Activating Token Index: 98

Text #7

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.7301. Min Act: -0.0000

Data Index: 1591958 (Python Code)

Max Activating Token Index: 138

Text #8

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.6914. Min Act: -0.0000

Data Index: 619238 (C4 (Web Text))

Max Activating Token Index: 649

Text #9

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.6927. Min Act: -0.0000

Data Index: 982036 (C4 (Web Text))

Max Activating Token Index: 663

Text #10

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.7209. Min Act: -0.0000

Data Index: 828275 (C4 (Web Text))

Max Activating Token Index: 947

Text #11

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.7192. Min Act: -0.0000

Data Index: 806394 (C4 (Web Text))

Max Activating Token Index: 747

Text #12

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.7213. Min Act: -0.0000

Data Index: 1313108 (C4 (Web Text))

Max Activating Token Index: 703

Text #13

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.6977. Min Act: -0.0000

Data Index: 1305863 (C4 (Web Text))

Max Activating Token Index: 301

Text #14

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.6622. Min Act: -0.0000

Data Index: 370485 (C4 (Web Text))

Max Activating Token Index: 496

Text #15

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.7132. Min Act: -0.0000

Data Index: 1070942 (C4 (Web Text))

Max Activating Token Index: 55

Text #16

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.6343. Min Act: -0.0000

Data Index: 487512 (C4 (Web Text))

Max Activating Token Index: 921

Text #17

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.6283. Min Act: -0.0000

Data Index: 252360 (C4 (Web Text))

Max Activating Token Index: 324

Text #18

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.6858. Min Act: -0.0000

Data Index: 1288271 (C4 (Web Text))

Max Activating Token Index: 560

Text #19

Max Range: 1.0522. Min Range: -1.0522

Max Act: 0.6417. Min Act: -0.0000

Data Index: 1391131 (Python Code)

Max Activating Token Index: 235