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

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

Neuron 166 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: 0.1798. Min Range: -0.1798

Max Act: 0.1625. Min Act: -0.0003

Data Index: 119382 (C4 (Web Text))

Max Activating Token Index: 967

Text #1

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1689. Min Act: -0.0003

Data Index: 1244154 (C4 (Web Text))

Max Activating Token Index: 695

Text #2

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1599. Min Act: -0.0003

Data Index: 1274914 (C4 (Web Text))

Max Activating Token Index: 749

Text #3

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1798. Min Act: -0.0003

Data Index: 93300 (C4 (Web Text))

Max Activating Token Index: 609

Text #4

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1662. Min Act: -0.0003

Data Index: 865762 (C4 (Web Text))

Max Activating Token Index: 773

Text #5

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1538. Min Act: -0.0003

Data Index: 1218965 (C4 (Web Text))

Max Activating Token Index: 788

Text #6

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1711. Min Act: -0.0003

Data Index: 1221616 (C4 (Web Text))

Max Activating Token Index: 746

Text #7

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1613. Min Act: -0.0004

Data Index: 589108 (C4 (Web Text))

Max Activating Token Index: 745

Text #8

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1615. Min Act: -0.0003

Data Index: 625072 (C4 (Web Text))

Max Activating Token Index: 313

Text #9

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1560. Min Act: -0.0003

Data Index: 334304 (C4 (Web Text))

Max Activating Token Index: 785

Text #10

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1588. Min Act: -0.0003

Data Index: 368051 (C4 (Web Text))

Max Activating Token Index: 303

Text #11

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1425. Min Act: -0.0003

Data Index: 349396 (C4 (Web Text))

Max Activating Token Index: 746

Text #12

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1468. Min Act: -0.0003

Data Index: 686615 (C4 (Web Text))

Max Activating Token Index: 771

Text #13

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1324. Min Act: -0.0004

Data Index: 1280575 (C4 (Web Text))

Max Activating Token Index: 977

Text #14

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1401. Min Act: -0.0003

Data Index: 310656 (C4 (Web Text))

Max Activating Token Index: 1016

Text #15

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1439. Min Act: -0.0003

Data Index: 1588739 (Python Code)

Max Activating Token Index: 858

Text #16

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1494. Min Act: -0.0003

Data Index: 14713 (C4 (Web Text))

Max Activating Token Index: 899

Text #17

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1490. Min Act: -0.0003

Data Index: 1542305 (Python Code)

Max Activating Token Index: 82

Text #18

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1435. Min Act: -0.0003

Data Index: 849392 (C4 (Web Text))

Max Activating Token Index: 787

Text #19

Max Range: 0.1798. Min Range: -0.1798

Max Act: 0.1531. Min Act: -0.0003

Data Index: 1494171 (Python Code)

Max Activating Token Index: 983