Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 1964 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.8657. Min Act: -0.0000

Data Index: 1468103 (Python Code)

Max Activating Token Index: 889

Text #1

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.7902. Min Act: -0.0001

Data Index: 464854 (C4 (Web Text))

Max Activating Token Index: 14

Text #2

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.7145. Min Act: -0.0000

Data Index: 1380719 (Python Code)

Max Activating Token Index: 926

Text #3

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6837. Min Act: -0.0001

Data Index: 126638 (C4 (Web Text))

Max Activating Token Index: 45

Text #4

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6492. Min Act: -0.0000

Data Index: 1361353 (Python Code)

Max Activating Token Index: 218

Text #5

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6671. Min Act: -0.0001

Data Index: 799263 (C4 (Web Text))

Max Activating Token Index: 68

Text #6

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6887. Min Act: -0.0001

Data Index: 770402 (C4 (Web Text))

Max Activating Token Index: 491

Text #7

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6242. Min Act: -0.0001

Data Index: 1307017 (C4 (Web Text))

Max Activating Token Index: 662

Text #8

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6683. Min Act: -0.0001

Data Index: 1151989 (C4 (Web Text))

Max Activating Token Index: 275

Text #9

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6359. Min Act: -0.0000

Data Index: 1579971 (Python Code)

Max Activating Token Index: 923

Text #10

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6606. Min Act: -0.0001

Data Index: 746110 (C4 (Web Text))

Max Activating Token Index: 366

Text #11

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6340. Min Act: -0.0000

Data Index: 1434930 (Python Code)

Max Activating Token Index: 706

Text #12

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6552. Min Act: -0.0001

Data Index: 55964 (C4 (Web Text))

Max Activating Token Index: 895

Text #13

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6499. Min Act: -0.0001

Data Index: 1554648 (Python Code)

Max Activating Token Index: 154

Text #14

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.5811. Min Act: -0.0001

Data Index: 226768 (C4 (Web Text))

Max Activating Token Index: 837

Text #15

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6088. Min Act: -0.0000

Data Index: 543751 (C4 (Web Text))

Max Activating Token Index: 463

Text #16

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.6115. Min Act: -0.0001

Data Index: 1653817 (Python Code)

Max Activating Token Index: 770

Text #17

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.5717. Min Act: -0.0000

Data Index: 1567209 (Python Code)

Max Activating Token Index: 347

Text #18

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.5807. Min Act: -0.0000

Data Index: 1368391 (Python Code)

Max Activating Token Index: 398

Text #19

Max Range: 0.8657. Min Range: -0.8657

Max Act: 0.5873. Min Act: -0.0001

Data Index: 722958 (C4 (Web Text))

Max Activating Token Index: 268