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

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

Neuron 250 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.1071. Min Act: -0.0001

Data Index: 94839 (C4 (Web Text))

Max Activating Token Index: 325

Text #1

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0932. Min Act: -0.0001

Data Index: 667708 (C4 (Web Text))

Max Activating Token Index: 156

Text #2

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0826. Min Act: -0.0001

Data Index: 233576 (C4 (Web Text))

Max Activating Token Index: 869

Text #3

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0819. Min Act: -0.0001

Data Index: 949282 (C4 (Web Text))

Max Activating Token Index: 463

Text #4

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0716. Min Act: -0.0001

Data Index: 413275 (C4 (Web Text))

Max Activating Token Index: 650

Text #5

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0808. Min Act: -0.0001

Data Index: 748362 (C4 (Web Text))

Max Activating Token Index: 791

Text #6

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0762. Min Act: -0.0001

Data Index: 1305432 (C4 (Web Text))

Max Activating Token Index: 664

Text #7

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0784. Min Act: -0.0001

Data Index: 1396435 (Python Code)

Max Activating Token Index: 545

Text #8

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0719. Min Act: -0.0001

Data Index: 121685 (C4 (Web Text))

Max Activating Token Index: 555

Text #9

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0720. Min Act: -0.0001

Data Index: 383079 (C4 (Web Text))

Max Activating Token Index: 475

Text #10

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0727. Min Act: -0.0001

Data Index: 1561149 (Python Code)

Max Activating Token Index: 545

Text #11

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0701. Min Act: -0.0001

Data Index: 503982 (C4 (Web Text))

Max Activating Token Index: 790

Text #12

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0672. Min Act: -0.0001

Data Index: 1445604 (Python Code)

Max Activating Token Index: 669

Text #13

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0659. Min Act: -0.0001

Data Index: 1507521 (Python Code)

Max Activating Token Index: 725

Text #14

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0655. Min Act: -0.0001

Data Index: 73399 (C4 (Web Text))

Max Activating Token Index: 442

Text #15

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0677. Min Act: -0.0001

Data Index: 138845 (C4 (Web Text))

Max Activating Token Index: 430

Text #16

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0663. Min Act: -0.0001

Data Index: 1308675 (C4 (Web Text))

Max Activating Token Index: 178

Text #17

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0697. Min Act: -0.0001

Data Index: 723057 (C4 (Web Text))

Max Activating Token Index: 500

Text #18

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0642. Min Act: -0.0001

Data Index: 1014135 (C4 (Web Text))

Max Activating Token Index: 35

Text #19

Max Range: 0.1071. Min Range: -0.1071

Max Act: 0.0716. Min Act: -0.0001

Data Index: 1151122 (C4 (Web Text))

Max Activating Token Index: 1011