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

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

Neuron 1453 in Layer 6

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.6658. Min Range: -3.6658

Max Act: 3.6658. Min Act: -0.0000

Data Index: 1242004 (C4 (Web Text))

Max Activating Token Index: 911

Text #1

Max Range: 3.6658. Min Range: -3.6658

Max Act: 3.6385. Min Act: -0.0000

Data Index: 1092605 (C4 (Web Text))

Max Activating Token Index: 909

Text #2

Max Range: 3.6658. Min Range: -3.6658

Max Act: 3.3791. Min Act: -0.0000

Data Index: 1482040 (Python Code)

Max Activating Token Index: 773

Text #3

Max Range: 3.6658. Min Range: -3.6658

Max Act: 3.5080. Min Act: -0.0000

Data Index: 688296 (C4 (Web Text))

Max Activating Token Index: 609

Text #4

Max Range: 3.6658. Min Range: -3.6658

Max Act: 3.1988. Min Act: -0.0000

Data Index: 90532 (C4 (Web Text))

Max Activating Token Index: 989

Text #5

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.9938. Min Act: -0.0000

Data Index: 772455 (C4 (Web Text))

Max Activating Token Index: 768

Text #6

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.9592. Min Act: -0.0000

Data Index: 1591820 (Python Code)

Max Activating Token Index: 642

Text #7

Max Range: 3.6658. Min Range: -3.6658

Max Act: 3.0045. Min Act: -0.0000

Data Index: 483819 (C4 (Web Text))

Max Activating Token Index: 169

Text #8

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.9958. Min Act: -0.0000

Data Index: 1399436 (Python Code)

Max Activating Token Index: 230

Text #9

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.8693. Min Act: -0.0000

Data Index: 1635952 (Python Code)

Max Activating Token Index: 106

Text #10

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.8912. Min Act: -0.0000

Data Index: 1378505 (Python Code)

Max Activating Token Index: 138

Text #11

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.7650. Min Act: -0.0000

Data Index: 1432012 (Python Code)

Max Activating Token Index: 992

Text #12

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.8457. Min Act: -0.0000

Data Index: 1190151 (C4 (Web Text))

Max Activating Token Index: 876

Text #13

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.9876. Min Act: -0.0000

Data Index: 1406825 (Python Code)

Max Activating Token Index: 430

Text #14

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.9667. Min Act: -0.0000

Data Index: 904103 (C4 (Web Text))

Max Activating Token Index: 98

Text #15

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.6962. Min Act: -0.0000

Data Index: 505296 (C4 (Web Text))

Max Activating Token Index: 132

Text #16

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.7040. Min Act: -0.0000

Data Index: 1197478 (C4 (Web Text))

Max Activating Token Index: 413

Text #17

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.8205. Min Act: -0.0000

Data Index: 136470 (C4 (Web Text))

Max Activating Token Index: 893

Text #18

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.8075. Min Act: -0.0000

Data Index: 966376 (C4 (Web Text))

Max Activating Token Index: 599

Text #19

Max Range: 3.6658. Min Range: -3.6658

Max Act: 2.7969. Min Act: -0.0000

Data Index: 1222909 (C4 (Web Text))

Max Activating Token Index: 392