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

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

Neuron 3958 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: 0.7314. Min Range: -0.7314

Max Act: 0.7314. Min Act: -0.0001

Data Index: 1526779 (Python Code)

Max Activating Token Index: 681

Text #1

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.6454. Min Act: -0.0001

Data Index: 1360885 (Python Code)

Max Activating Token Index: 35

Text #2

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.5954. Min Act: -0.0001

Data Index: 342667 (C4 (Web Text))

Max Activating Token Index: 712

Text #3

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.6426. Min Act: -0.0001

Data Index: 1483633 (Python Code)

Max Activating Token Index: 854

Text #4

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.5754. Min Act: -0.0000

Data Index: 1544561 (Python Code)

Max Activating Token Index: 600

Text #5

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.5010. Min Act: -0.0000

Data Index: 1500260 (Python Code)

Max Activating Token Index: 471

Text #6

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4953. Min Act: -0.0001

Data Index: 1632189 (Python Code)

Max Activating Token Index: 813

Text #7

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.5127. Min Act: -0.0001

Data Index: 1411270 (Python Code)

Max Activating Token Index: 97

Text #8

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.5161. Min Act: -0.0001

Data Index: 1534050 (Python Code)

Max Activating Token Index: 654

Text #9

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4649. Min Act: -0.0000

Data Index: 1459301 (Python Code)

Max Activating Token Index: 755

Text #10

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4705. Min Act: -0.0001

Data Index: 1608156 (Python Code)

Max Activating Token Index: 462

Text #11

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4954. Min Act: -0.0001

Data Index: 1071136 (C4 (Web Text))

Max Activating Token Index: 123

Text #12

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4675. Min Act: -0.0001

Data Index: 1384019 (Python Code)

Max Activating Token Index: 592

Text #13

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4218. Min Act: -0.0001

Data Index: 714589 (C4 (Web Text))

Max Activating Token Index: 272

Text #14

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4621. Min Act: -0.0001

Data Index: 1599360 (Python Code)

Max Activating Token Index: 484

Text #15

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4431. Min Act: -0.0001

Data Index: 1415003 (Python Code)

Max Activating Token Index: 170

Text #16

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4113. Min Act: -0.0001

Data Index: 605019 (C4 (Web Text))

Max Activating Token Index: 918

Text #17

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4240. Min Act: -0.0001

Data Index: 344267 (C4 (Web Text))

Max Activating Token Index: 310

Text #18

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4413. Min Act: -0.0001

Data Index: 1550827 (Python Code)

Max Activating Token Index: 371

Text #19

Max Range: 0.7314. Min Range: -0.7314

Max Act: 0.4151. Min Act: -0.0001

Data Index: 1446675 (Python Code)

Max Activating Token Index: 266