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

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

Neuron 4281 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.9078. Min Act: -0.0000

Data Index: 70562 (C4 (Web Text))

Max Activating Token Index: 568

Text #1

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.8676. Min Act: -0.0000

Data Index: 531288 (C4 (Web Text))

Max Activating Token Index: 134

Text #2

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.8848. Min Act: -0.0000

Data Index: 1217993 (C4 (Web Text))

Max Activating Token Index: 57

Text #3

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.8368. Min Act: -0.0000

Data Index: 679306 (C4 (Web Text))

Max Activating Token Index: 452

Text #4

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.8138. Min Act: -0.0000

Data Index: 500284 (C4 (Web Text))

Max Activating Token Index: 822

Text #5

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.8404. Min Act: -0.0000

Data Index: 1610105 (Python Code)

Max Activating Token Index: 623

Text #6

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.6647. Min Act: -0.0000

Data Index: 961469 (C4 (Web Text))

Max Activating Token Index: 607

Text #7

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.6966. Min Act: -0.0000

Data Index: 1650053 (Python Code)

Max Activating Token Index: 708

Text #8

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.7830. Min Act: -0.0000

Data Index: 960248 (C4 (Web Text))

Max Activating Token Index: 581

Text #9

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.8137. Min Act: -0.0000

Data Index: 595899 (C4 (Web Text))

Max Activating Token Index: 652

Text #10

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.7635. Min Act: -0.0000

Data Index: 552580 (C4 (Web Text))

Max Activating Token Index: 517

Text #11

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.7175. Min Act: -0.0000

Data Index: 1371831 (Python Code)

Max Activating Token Index: 553

Text #12

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.7846. Min Act: -0.0000

Data Index: 260446 (C4 (Web Text))

Max Activating Token Index: 743

Text #13

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.6885. Min Act: -0.0000

Data Index: 361031 (C4 (Web Text))

Max Activating Token Index: 459

Text #14

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.7369. Min Act: -0.0000

Data Index: 1064953 (C4 (Web Text))

Max Activating Token Index: 615

Text #15

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.6199. Min Act: -0.0000

Data Index: 1012693 (C4 (Web Text))

Max Activating Token Index: 802

Text #16

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.8089. Min Act: -0.0000

Data Index: 776660 (C4 (Web Text))

Max Activating Token Index: 161

Text #17

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.8278. Min Act: -0.0000

Data Index: 604593 (C4 (Web Text))

Max Activating Token Index: 662

Text #18

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.7153. Min Act: -0.0000

Data Index: 610242 (C4 (Web Text))

Max Activating Token Index: 907

Text #19

Max Range: 0.9078. Min Range: -0.9078

Max Act: 0.6599. Min Act: -0.0000

Data Index: 373572 (C4 (Web Text))

Max Activating Token Index: 688