Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 3320 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.8545. Min Act: -0.0002

Data Index: 1167851 (C4 (Web Text))

Max Activating Token Index: 774

Text #1

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.7625. Min Act: -0.0002

Data Index: 1297841 (C4 (Web Text))

Max Activating Token Index: 546

Text #2

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.8363. Min Act: -0.0002

Data Index: 1231836 (C4 (Web Text))

Max Activating Token Index: 762

Text #3

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6901. Min Act: -0.0002

Data Index: 678764 (C4 (Web Text))

Max Activating Token Index: 902

Text #4

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.7088. Min Act: -0.0002

Data Index: 891830 (C4 (Web Text))

Max Activating Token Index: 232

Text #5

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6856. Min Act: -0.0001

Data Index: 164332 (C4 (Web Text))

Max Activating Token Index: 499

Text #6

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6981. Min Act: -0.0002

Data Index: 704105 (C4 (Web Text))

Max Activating Token Index: 904

Text #7

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.7152. Min Act: -0.0001

Data Index: 208396 (C4 (Web Text))

Max Activating Token Index: 875

Text #8

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6665. Min Act: -0.0002

Data Index: 524639 (C4 (Web Text))

Max Activating Token Index: 408

Text #9

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.7217. Min Act: -0.0001

Data Index: 702774 (C4 (Web Text))

Max Activating Token Index: 897

Text #10

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6324. Min Act: -0.0002

Data Index: 852115 (C4 (Web Text))

Max Activating Token Index: 933

Text #11

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6647. Min Act: -0.0001

Data Index: 1380552 (Python Code)

Max Activating Token Index: 933

Text #12

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6585. Min Act: -0.0002

Data Index: 1003930 (C4 (Web Text))

Max Activating Token Index: 464

Text #13

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6675. Min Act: -0.0002

Data Index: 910459 (C4 (Web Text))

Max Activating Token Index: 473

Text #14

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6429. Min Act: -0.0002

Data Index: 863048 (C4 (Web Text))

Max Activating Token Index: 532

Text #15

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6050. Min Act: -0.0001

Data Index: 1057592 (C4 (Web Text))

Max Activating Token Index: 332

Text #16

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6255. Min Act: -0.0001

Data Index: 1020275 (C4 (Web Text))

Max Activating Token Index: 826

Text #17

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6223. Min Act: -0.0002

Data Index: 45599 (C4 (Web Text))

Max Activating Token Index: 634

Text #18

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6323. Min Act: -0.0002

Data Index: 992927 (C4 (Web Text))

Max Activating Token Index: 886

Text #19

Max Range: 0.8545. Min Range: -0.8545

Max Act: 0.6686. Min Act: -0.0002

Data Index: 678800 (C4 (Web Text))

Max Activating Token Index: 965