Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 201 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.4356. Min Act: -0.0002

Data Index: 1433025 (Python Code)

Max Activating Token Index: 17

Text #1

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3883. Min Act: -0.0002

Data Index: 858095 (C4 (Web Text))

Max Activating Token Index: 701

Text #2

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.4070. Min Act: -0.0002

Data Index: 1085554 (C4 (Web Text))

Max Activating Token Index: 442

Text #3

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3907. Min Act: -0.0002

Data Index: 1070451 (C4 (Web Text))

Max Activating Token Index: 26

Text #4

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.4102. Min Act: -0.0002

Data Index: 453640 (C4 (Web Text))

Max Activating Token Index: 203

Text #5

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3924. Min Act: -0.0002

Data Index: 887937 (C4 (Web Text))

Max Activating Token Index: 622

Text #6

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3523. Min Act: -0.0002

Data Index: 297487 (C4 (Web Text))

Max Activating Token Index: 317

Text #7

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3816. Min Act: -0.0002

Data Index: 915021 (C4 (Web Text))

Max Activating Token Index: 259

Text #8

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3744. Min Act: -0.0002

Data Index: 1149005 (C4 (Web Text))

Max Activating Token Index: 517

Text #9

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3901. Min Act: -0.0002

Data Index: 800875 (C4 (Web Text))

Max Activating Token Index: 642

Text #10

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3624. Min Act: -0.0002

Data Index: 407350 (C4 (Web Text))

Max Activating Token Index: 338

Text #11

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3767. Min Act: -0.0002

Data Index: 1388574 (Python Code)

Max Activating Token Index: 672

Text #12

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3787. Min Act: -0.0002

Data Index: 366725 (C4 (Web Text))

Max Activating Token Index: 752

Text #13

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3817. Min Act: -0.0002

Data Index: 1264872 (C4 (Web Text))

Max Activating Token Index: 600

Text #14

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3480. Min Act: -0.0002

Data Index: 579726 (C4 (Web Text))

Max Activating Token Index: 980

Text #15

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3491. Min Act: -0.0002

Data Index: 412732 (C4 (Web Text))

Max Activating Token Index: 567

Text #16

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3884. Min Act: -0.0002

Data Index: 17915 (C4 (Web Text))

Max Activating Token Index: 810

Text #17

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3671. Min Act: -0.0002

Data Index: 1521540 (Python Code)

Max Activating Token Index: 23

Text #18

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3533. Min Act: -0.0002

Data Index: 1158227 (C4 (Web Text))

Max Activating Token Index: 155

Text #19

Max Range: 0.4356. Min Range: -0.4356

Max Act: 0.3606. Min Act: -0.0002

Data Index: 66674 (C4 (Web Text))

Max Activating Token Index: 440