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

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

Neuron 4945 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.0824. Min Range: -1.0824

Max Act: 1.0824. Min Act: -0.0001

Data Index: 634021 (C4 (Web Text))

Max Activating Token Index: 799

Text #1

Max Range: 1.0824. Min Range: -1.0824

Max Act: 1.0642. Min Act: -0.0001

Data Index: 613496 (C4 (Web Text))

Max Activating Token Index: 20

Text #2

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8368. Min Act: -0.0001

Data Index: 729942 (C4 (Web Text))

Max Activating Token Index: 847

Text #3

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8734. Min Act: -0.0001

Data Index: 695147 (C4 (Web Text))

Max Activating Token Index: 62

Text #4

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8640. Min Act: -0.0001

Data Index: 899087 (C4 (Web Text))

Max Activating Token Index: 916

Text #5

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8650. Min Act: -0.0001

Data Index: 1523079 (Python Code)

Max Activating Token Index: 154

Text #6

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8605. Min Act: -0.0001

Data Index: 1587794 (Python Code)

Max Activating Token Index: 193

Text #7

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.9217. Min Act: -0.0001

Data Index: 1572349 (Python Code)

Max Activating Token Index: 26

Text #8

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.7852. Min Act: -0.0001

Data Index: 1167146 (C4 (Web Text))

Max Activating Token Index: 313

Text #9

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8687. Min Act: -0.0001

Data Index: 358444 (C4 (Web Text))

Max Activating Token Index: 922

Text #10

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.9052. Min Act: -0.0001

Data Index: 89382 (C4 (Web Text))

Max Activating Token Index: 158

Text #11

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8191. Min Act: -0.0001

Data Index: 31944 (C4 (Web Text))

Max Activating Token Index: 528

Text #12

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8205. Min Act: -0.0001

Data Index: 270503 (C4 (Web Text))

Max Activating Token Index: 336

Text #13

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8897. Min Act: -0.0001

Data Index: 760681 (C4 (Web Text))

Max Activating Token Index: 160

Text #14

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8915. Min Act: -0.0001

Data Index: 489698 (C4 (Web Text))

Max Activating Token Index: 660

Text #15

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8854. Min Act: -0.0001

Data Index: 188929 (C4 (Web Text))

Max Activating Token Index: 830

Text #16

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8043. Min Act: -0.0001

Data Index: 1393322 (Python Code)

Max Activating Token Index: 36

Text #17

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.8302. Min Act: -0.0001

Data Index: 1018454 (C4 (Web Text))

Max Activating Token Index: 230

Text #18

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.7481. Min Act: -0.0001

Data Index: 683393 (C4 (Web Text))

Max Activating Token Index: 613

Text #19

Max Range: 1.0824. Min Range: -1.0824

Max Act: 0.7778. Min Act: -0.0001

Data Index: 291604 (C4 (Web Text))

Max Activating Token Index: 245