Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 651 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 3.4757. Min Range: -3.4757

Max Act: 3.4757. Min Act: -0.0000

Data Index: 183819 (C4 (Web Text))

Max Activating Token Index: 905

Text #1

Max Range: 3.4757. Min Range: -3.4757

Max Act: 3.3818. Min Act: -0.0000

Data Index: 1033201 (C4 (Web Text))

Max Activating Token Index: 87

Text #2

Max Range: 3.4757. Min Range: -3.4757

Max Act: 3.1458. Min Act: -0.0000

Data Index: 994660 (C4 (Web Text))

Max Activating Token Index: 1017

Text #3

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.9632. Min Act: -0.0000

Data Index: 1073510 (C4 (Web Text))

Max Activating Token Index: 588

Text #4

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.7377. Min Act: -0.0000

Data Index: 1417435 (Python Code)

Max Activating Token Index: 738

Text #5

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.6109. Min Act: -0.0000

Data Index: 1260087 (C4 (Web Text))

Max Activating Token Index: 712

Text #6

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.5793. Min Act: -0.0000

Data Index: 633765 (C4 (Web Text))

Max Activating Token Index: 150

Text #7

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.4315. Min Act: -0.0000

Data Index: 145462 (C4 (Web Text))

Max Activating Token Index: 521

Text #8

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.5059. Min Act: -0.0000

Data Index: 1384941 (Python Code)

Max Activating Token Index: 538

Text #9

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.4615. Min Act: -0.0000

Data Index: 1562242 (Python Code)

Max Activating Token Index: 590

Text #10

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.2102. Min Act: -0.0000

Data Index: 308144 (C4 (Web Text))

Max Activating Token Index: 798

Text #11

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.0644. Min Act: -0.0000

Data Index: 1043677 (C4 (Web Text))

Max Activating Token Index: 355

Text #12

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.6285. Min Act: -0.0000

Data Index: 623666 (C4 (Web Text))

Max Activating Token Index: 137

Text #13

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.5031. Min Act: -0.0000

Data Index: 1634774 (Python Code)

Max Activating Token Index: 798

Text #14

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.1449. Min Act: -0.0000

Data Index: 889767 (C4 (Web Text))

Max Activating Token Index: 289

Text #15

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.4045. Min Act: -0.0000

Data Index: 1536933 (Python Code)

Max Activating Token Index: 426

Text #16

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.2085. Min Act: -0.0000

Data Index: 1508632 (Python Code)

Max Activating Token Index: 907

Text #17

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.2144. Min Act: -0.0000

Data Index: 693114 (C4 (Web Text))

Max Activating Token Index: 201

Text #18

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.3015. Min Act: -0.0000

Data Index: 1436207 (Python Code)

Max Activating Token Index: 369

Text #19

Max Range: 3.4757. Min Range: -3.4757

Max Act: 2.1942. Min Act: -0.0000

Data Index: 910835 (C4 (Web Text))

Max Activating Token Index: 220