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

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

Neuron 882 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.4334. Min Act: -0.0001

Data Index: 306196 (C4 (Web Text))

Max Activating Token Index: 802

Text #1

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.3868. Min Act: -0.0001

Data Index: 1646552 (Python Code)

Max Activating Token Index: 841

Text #2

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.1779. Min Act: -0.0001

Data Index: 236985 (C4 (Web Text))

Max Activating Token Index: 484

Text #3

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.1645. Min Act: -0.0001

Data Index: 1261545 (C4 (Web Text))

Max Activating Token Index: 132

Text #4

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.1541. Min Act: -0.0001

Data Index: 733972 (C4 (Web Text))

Max Activating Token Index: 147

Text #5

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.0822. Min Act: -0.0001

Data Index: 900839 (C4 (Web Text))

Max Activating Token Index: 112

Text #6

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.1802. Min Act: -0.0001

Data Index: 1393064 (Python Code)

Max Activating Token Index: 521

Text #7

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.0254. Min Act: -0.0001

Data Index: 824591 (C4 (Web Text))

Max Activating Token Index: 662

Text #8

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.1515. Min Act: -0.0001

Data Index: 690659 (C4 (Web Text))

Max Activating Token Index: 430

Text #9

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.1800. Min Act: -0.0001

Data Index: 885070 (C4 (Web Text))

Max Activating Token Index: 672

Text #10

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.0723. Min Act: -0.0001

Data Index: 454956 (C4 (Web Text))

Max Activating Token Index: 90

Text #11

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.0411. Min Act: -0.0001

Data Index: 1623727 (Python Code)

Max Activating Token Index: 107

Text #12

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.0517. Min Act: -0.0001

Data Index: 98358 (C4 (Web Text))

Max Activating Token Index: 761

Text #13

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.0228. Min Act: -0.0001

Data Index: 242015 (C4 (Web Text))

Max Activating Token Index: 824

Text #14

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.0147. Min Act: -0.0001

Data Index: 87686 (C4 (Web Text))

Max Activating Token Index: 327

Text #15

Max Range: 1.4334. Min Range: -1.4334

Max Act: 1.0448. Min Act: -0.0001

Data Index: 88104 (C4 (Web Text))

Max Activating Token Index: 132

Text #16

Max Range: 1.4334. Min Range: -1.4334

Max Act: 0.9232. Min Act: -0.0001

Data Index: 685366 (C4 (Web Text))

Max Activating Token Index: 952

Text #17

Max Range: 1.4334. Min Range: -1.4334

Max Act: 0.9760. Min Act: -0.0001

Data Index: 1408753 (Python Code)

Max Activating Token Index: 20

Text #18

Max Range: 1.4334. Min Range: -1.4334

Max Act: 0.9603. Min Act: -0.0001

Data Index: 135869 (C4 (Web Text))

Max Activating Token Index: 869

Text #19

Max Range: 1.4334. Min Range: -1.4334

Max Act: 0.9484. Min Act: -0.0001

Data Index: 192405 (C4 (Web Text))

Max Activating Token Index: 354