Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 1250 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.5348. Min Act: -0.0003

Data Index: 1228362 (C4 (Web Text))

Max Activating Token Index: 244

Text #1

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.4518. Min Act: -0.0003

Data Index: 432171 (C4 (Web Text))

Max Activating Token Index: 655

Text #2

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.4110. Min Act: -0.0003

Data Index: 849489 (C4 (Web Text))

Max Activating Token Index: 190

Text #3

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.4620. Min Act: -0.0003

Data Index: 1067430 (C4 (Web Text))

Max Activating Token Index: 721

Text #4

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3207. Min Act: -0.0003

Data Index: 576030 (C4 (Web Text))

Max Activating Token Index: 301

Text #5

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.5375. Min Act: -0.0003

Data Index: 1644466 (Python Code)

Max Activating Token Index: 316

Text #6

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3941. Min Act: -0.0003

Data Index: 1188003 (C4 (Web Text))

Max Activating Token Index: 932

Text #7

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3892. Min Act: -0.0003

Data Index: 1032626 (C4 (Web Text))

Max Activating Token Index: 51

Text #8

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3964. Min Act: -0.0003

Data Index: 1039610 (C4 (Web Text))

Max Activating Token Index: 415

Text #9

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3332. Min Act: -0.0003

Data Index: 798604 (C4 (Web Text))

Max Activating Token Index: 291

Text #10

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3491. Min Act: -0.0003

Data Index: 1090375 (C4 (Web Text))

Max Activating Token Index: 120

Text #11

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3775. Min Act: -0.0003

Data Index: 733576 (C4 (Web Text))

Max Activating Token Index: 898

Text #12

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3695. Min Act: -0.0003

Data Index: 648791 (C4 (Web Text))

Max Activating Token Index: 842

Text #13

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3589. Min Act: -0.0003

Data Index: 1655227 (Python Code)

Max Activating Token Index: 700

Text #14

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.2719. Min Act: -0.0003

Data Index: 195034 (C4 (Web Text))

Max Activating Token Index: 269

Text #15

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3525. Min Act: -0.0003

Data Index: 1425854 (Python Code)

Max Activating Token Index: 972

Text #16

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3079. Min Act: -0.0003

Data Index: 1331115 (C4 (Web Text))

Max Activating Token Index: 801

Text #17

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3521. Min Act: -0.0003

Data Index: 917220 (C4 (Web Text))

Max Activating Token Index: 178

Text #18

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3596. Min Act: -0.0003

Data Index: 644285 (C4 (Web Text))

Max Activating Token Index: 389

Text #19

Max Range: 0.5375. Min Range: -0.5375

Max Act: 0.3963. Min Act: -0.0003

Data Index: 242306 (C4 (Web Text))

Max Activating Token Index: 672