Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 621 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 10.1767. Min Range: -10.1767

Max Act: 10.1767. Min Act: -0.0005

Data Index: 1237746 (C4 (Web Text))

Max Activating Token Index: 710

Text #1

Max Range: 10.1767. Min Range: -10.1767

Max Act: 9.7841. Min Act: -0.0005

Data Index: 1612982 (Python Code)

Max Activating Token Index: 963

Text #2

Max Range: 10.1767. Min Range: -10.1767

Max Act: 9.4096. Min Act: -0.0005

Data Index: 249978 (C4 (Web Text))

Max Activating Token Index: 903

Text #3

Max Range: 10.1767. Min Range: -10.1767

Max Act: 9.4017. Min Act: -0.0005

Data Index: 510425 (C4 (Web Text))

Max Activating Token Index: 964

Text #4

Max Range: 10.1767. Min Range: -10.1767

Max Act: 8.9301. Min Act: -0.0005

Data Index: 739829 (C4 (Web Text))

Max Activating Token Index: 1023

Text #5

Max Range: 10.1767. Min Range: -10.1767

Max Act: 7.2189. Min Act: -0.0005

Data Index: 478991 (C4 (Web Text))

Max Activating Token Index: 1009

Text #6

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.8663. Min Act: -0.0005

Data Index: 864861 (C4 (Web Text))

Max Activating Token Index: 1023

Text #7

Max Range: 10.1767. Min Range: -10.1767

Max Act: 7.2834. Min Act: -0.0005

Data Index: 898982 (C4 (Web Text))

Max Activating Token Index: 952

Text #8

Max Range: 10.1767. Min Range: -10.1767

Max Act: 7.2419. Min Act: -0.0005

Data Index: 433739 (C4 (Web Text))

Max Activating Token Index: 1023

Text #9

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.7646. Min Act: -0.0005

Data Index: 72924 (C4 (Web Text))

Max Activating Token Index: 1023

Text #10

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.5488. Min Act: -0.0005

Data Index: 550032 (C4 (Web Text))

Max Activating Token Index: 1023

Text #11

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.4481. Min Act: -0.0005

Data Index: 1632648 (Python Code)

Max Activating Token Index: 356

Text #12

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.5525. Min Act: -0.0005

Data Index: 679134 (C4 (Web Text))

Max Activating Token Index: 1023

Text #13

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.2503. Min Act: -0.0005

Data Index: 609196 (C4 (Web Text))

Max Activating Token Index: 1023

Text #14

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.9119. Min Act: -0.0005

Data Index: 204652 (C4 (Web Text))

Max Activating Token Index: 1023

Text #15

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.5237. Min Act: -0.0005

Data Index: 224272 (C4 (Web Text))

Max Activating Token Index: 1023

Text #16

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.1584. Min Act: -0.0005

Data Index: 78675 (C4 (Web Text))

Max Activating Token Index: 1023

Text #17

Max Range: 10.1767. Min Range: -10.1767

Max Act: 5.9033. Min Act: -0.0005

Data Index: 761422 (C4 (Web Text))

Max Activating Token Index: 943

Text #18

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.3417. Min Act: -0.0005

Data Index: 1175998 (C4 (Web Text))

Max Activating Token Index: 1023

Text #19

Max Range: 10.1767. Min Range: -10.1767

Max Act: 6.4920. Min Act: -0.0005

Data Index: 945404 (C4 (Web Text))

Max Activating Token Index: 1023