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

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

Neuron 808 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: 0.0812. Min Range: -0.0812

Max Act: 0.0812. Min Act: -0.0003

Data Index: 585116 (C4 (Web Text))

Max Activating Token Index: 851

Text #1

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0760. Min Act: -0.0003

Data Index: 1333630 (C4 (Web Text))

Max Activating Token Index: 778

Text #2

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0756. Min Act: -0.0003

Data Index: 1378124 (Python Code)

Max Activating Token Index: 582

Text #3

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0787. Min Act: -0.0003

Data Index: 244453 (C4 (Web Text))

Max Activating Token Index: 149

Text #4

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0715. Min Act: -0.0003

Data Index: 860185 (C4 (Web Text))

Max Activating Token Index: 912

Text #5

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0702. Min Act: -0.0003

Data Index: 91505 (C4 (Web Text))

Max Activating Token Index: 480

Text #6

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0695. Min Act: -0.0003

Data Index: 41115 (C4 (Web Text))

Max Activating Token Index: 279

Text #7

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0687. Min Act: -0.0003

Data Index: 1114269 (C4 (Web Text))

Max Activating Token Index: 666

Text #8

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0691. Min Act: -0.0004

Data Index: 1031492 (C4 (Web Text))

Max Activating Token Index: 226

Text #9

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0697. Min Act: -0.0003

Data Index: 1553863 (Python Code)

Max Activating Token Index: 84

Text #10

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0696. Min Act: -0.0003

Data Index: 551522 (C4 (Web Text))

Max Activating Token Index: 858

Text #11

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0699. Min Act: -0.0004

Data Index: 419622 (C4 (Web Text))

Max Activating Token Index: 429

Text #12

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0683. Min Act: -0.0003

Data Index: 878245 (C4 (Web Text))

Max Activating Token Index: 616

Text #13

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0678. Min Act: -0.0003

Data Index: 56812 (C4 (Web Text))

Max Activating Token Index: 738

Text #14

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0703. Min Act: -0.0003

Data Index: 625316 (C4 (Web Text))

Max Activating Token Index: 112

Text #15

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0711. Min Act: -0.0003

Data Index: 1151961 (C4 (Web Text))

Max Activating Token Index: 786

Text #16

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0677. Min Act: -0.0003

Data Index: 1359712 (C4 (Web Text))

Max Activating Token Index: 265

Text #17

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0660. Min Act: -0.0003

Data Index: 1543795 (Python Code)

Max Activating Token Index: 527

Text #18

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0655. Min Act: -0.0003

Data Index: 426976 (C4 (Web Text))

Max Activating Token Index: 908

Text #19

Max Range: 0.0812. Min Range: -0.0812

Max Act: 0.0685. Min Act: -0.0003

Data Index: 1437731 (Python Code)

Max Activating Token Index: 628