Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 314 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0530. Min Act: -0.0003

Data Index: 747157 (C4 (Web Text))

Max Activating Token Index: 183

Text #1

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0521. Min Act: -0.0001

Data Index: 884891 (C4 (Web Text))

Max Activating Token Index: 417

Text #2

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0507. Min Act: -0.0003

Data Index: 979060 (C4 (Web Text))

Max Activating Token Index: 236

Text #3

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0487. Min Act: -0.0003

Data Index: 391648 (C4 (Web Text))

Max Activating Token Index: 456

Text #4

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0496. Min Act: -0.0003

Data Index: 818787 (C4 (Web Text))

Max Activating Token Index: 479

Text #5

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0475. Min Act: -0.0003

Data Index: 1040567 (C4 (Web Text))

Max Activating Token Index: 894

Text #6

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0485. Min Act: -0.0004

Data Index: 314952 (C4 (Web Text))

Max Activating Token Index: 541

Text #7

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0475. Min Act: -0.0003

Data Index: 840541 (C4 (Web Text))

Max Activating Token Index: 601

Text #8

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0453. Min Act: -0.0002

Data Index: 926704 (C4 (Web Text))

Max Activating Token Index: 295

Text #9

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0449. Min Act: -0.0004

Data Index: 860414 (C4 (Web Text))

Max Activating Token Index: 102

Text #10

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0443. Min Act: -0.0002

Data Index: 164580 (C4 (Web Text))

Max Activating Token Index: 266

Text #11

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0446. Min Act: -0.0002

Data Index: 1125061 (C4 (Web Text))

Max Activating Token Index: 1014

Text #12

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0447. Min Act: -0.0002

Data Index: 402475 (C4 (Web Text))

Max Activating Token Index: 807

Text #13

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0430. Min Act: -0.0004

Data Index: 1639608 (Python Code)

Max Activating Token Index: 991

Text #14

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0438. Min Act: -0.0003

Data Index: 1012 (C4 (Web Text))

Max Activating Token Index: 575

Text #15

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0438. Min Act: -0.0002

Data Index: 1161804 (C4 (Web Text))

Max Activating Token Index: 1018

Text #16

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0433. Min Act: -0.0003

Data Index: 396408 (C4 (Web Text))

Max Activating Token Index: 543

Text #17

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0425. Min Act: -0.0003

Data Index: 905024 (C4 (Web Text))

Max Activating Token Index: 620

Text #18

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0418. Min Act: -0.0003

Data Index: 652615 (C4 (Web Text))

Max Activating Token Index: 923

Text #19

Max Range: 0.0530. Min Range: -0.0530

Max Act: 0.0422. Min Act: -0.0003

Data Index: 1341957 (C4 (Web Text))

Max Activating Token Index: 942