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

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

Neuron 4801 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.6068. Min Act: -0.0000

Data Index: 46548 (C4 (Web Text))

Max Activating Token Index: 952

Text #1

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.5372. Min Act: -0.0000

Data Index: 737658 (C4 (Web Text))

Max Activating Token Index: 430

Text #2

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4820. Min Act: -0.0000

Data Index: 200413 (C4 (Web Text))

Max Activating Token Index: 902

Text #3

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.5422. Min Act: -0.0000

Data Index: 770718 (C4 (Web Text))

Max Activating Token Index: 856

Text #4

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.5102. Min Act: -0.0000

Data Index: 229608 (C4 (Web Text))

Max Activating Token Index: 637

Text #5

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4915. Min Act: -0.0000

Data Index: 515161 (C4 (Web Text))

Max Activating Token Index: 192

Text #6

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4580. Min Act: -0.0000

Data Index: 257237 (C4 (Web Text))

Max Activating Token Index: 961

Text #7

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4987. Min Act: -0.0000

Data Index: 1596582 (Python Code)

Max Activating Token Index: 790

Text #8

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4529. Min Act: -0.0000

Data Index: 1405029 (Python Code)

Max Activating Token Index: 239

Text #9

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4599. Min Act: -0.0000

Data Index: 1632017 (Python Code)

Max Activating Token Index: 1002

Text #10

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4672. Min Act: -0.0000

Data Index: 266512 (C4 (Web Text))

Max Activating Token Index: 834

Text #11

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4137. Min Act: -0.0000

Data Index: 826388 (C4 (Web Text))

Max Activating Token Index: 535

Text #12

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.3945. Min Act: -0.0000

Data Index: 1575587 (Python Code)

Max Activating Token Index: 238

Text #13

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4659. Min Act: -0.0000

Data Index: 191350 (C4 (Web Text))

Max Activating Token Index: 650

Text #14

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4393. Min Act: -0.0000

Data Index: 745943 (C4 (Web Text))

Max Activating Token Index: 369

Text #15

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4167. Min Act: -0.0000

Data Index: 143650 (C4 (Web Text))

Max Activating Token Index: 982

Text #16

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4515. Min Act: -0.0000

Data Index: 1440511 (Python Code)

Max Activating Token Index: 827

Text #17

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4257. Min Act: -0.0000

Data Index: 1295480 (C4 (Web Text))

Max Activating Token Index: 352

Text #18

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4100. Min Act: -0.0000

Data Index: 1489754 (Python Code)

Max Activating Token Index: 292

Text #19

Max Range: 0.6068. Min Range: -0.6068

Max Act: 0.4368. Min Act: -0.0000

Data Index: 49076 (C4 (Web Text))

Max Activating Token Index: 534