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

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

Neuron 630 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0851. Min Act: -0.0002

Data Index: 292040 (C4 (Web Text))

Max Activating Token Index: 14

Text #1

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0764. Min Act: -0.0002

Data Index: 1602444 (Python Code)

Max Activating Token Index: 791

Text #2

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0819. Min Act: -0.0003

Data Index: 1351912 (C4 (Web Text))

Max Activating Token Index: 46

Text #3

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0808. Min Act: -0.0002

Data Index: 1409254 (Python Code)

Max Activating Token Index: 5

Text #4

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0808. Min Act: -0.0002

Data Index: 257959 (C4 (Web Text))

Max Activating Token Index: 5

Text #5

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0621. Min Act: -0.0002

Data Index: 1436692 (Python Code)

Max Activating Token Index: 280

Text #6

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0684. Min Act: -0.0002

Data Index: 1341498 (C4 (Web Text))

Max Activating Token Index: 6

Text #7

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0658. Min Act: -0.0003

Data Index: 948135 (C4 (Web Text))

Max Activating Token Index: 56

Text #8

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0860. Min Act: -0.0002

Data Index: 1487408 (Python Code)

Max Activating Token Index: 923

Text #9

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0828. Min Act: -0.0003

Data Index: 512755 (C4 (Web Text))

Max Activating Token Index: 78

Text #10

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0626. Min Act: -0.0002

Data Index: 1049071 (C4 (Web Text))

Max Activating Token Index: 689

Text #11

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0745. Min Act: -0.0002

Data Index: 818410 (C4 (Web Text))

Max Activating Token Index: 993

Text #12

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0620. Min Act: -0.0002

Data Index: 1413254 (Python Code)

Max Activating Token Index: 838

Text #13

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0615. Min Act: -0.0003

Data Index: 1178247 (C4 (Web Text))

Max Activating Token Index: 154

Text #14

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0806. Min Act: -0.0002

Data Index: 41794 (C4 (Web Text))

Max Activating Token Index: 165

Text #15

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0699. Min Act: -0.0002

Data Index: 101498 (C4 (Web Text))

Max Activating Token Index: 378

Text #16

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0779. Min Act: -0.0003

Data Index: 799316 (C4 (Web Text))

Max Activating Token Index: 388

Text #17

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0546. Min Act: -0.0002

Data Index: 651052 (C4 (Web Text))

Max Activating Token Index: 735

Text #18

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0753. Min Act: -0.0002

Data Index: 58023 (C4 (Web Text))

Max Activating Token Index: 859

Text #19

Max Range: 0.0860. Min Range: -0.0860

Max Act: 0.0601. Min Act: -0.0003

Data Index: 563446 (C4 (Web Text))

Max Activating Token Index: 284