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

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

Neuron 48 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.2659. Min Act: -0.1700

Data Index: 677401 (C4 (Web Text))

Max Activating Token Index: 338

Text #1

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.2188. Min Act: -0.1700

Data Index: 1427176 (Python Code)

Max Activating Token Index: 659

Text #2

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.2030. Min Act: -0.1700

Data Index: 1521554 (Python Code)

Max Activating Token Index: 536

Text #3

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.2030. Min Act: -0.1700

Data Index: 807293 (C4 (Web Text))

Max Activating Token Index: 694

Text #4

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.2030. Min Act: -0.1700

Data Index: 1649718 (Python Code)

Max Activating Token Index: 777

Text #5

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1715. Min Act: -0.1700

Data Index: 1584280 (Python Code)

Max Activating Token Index: 194

Text #6

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1715. Min Act: -0.1700

Data Index: 1432686 (Python Code)

Max Activating Token Index: 194

Text #7

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1873. Min Act: -0.1700

Data Index: 1410204 (Python Code)

Max Activating Token Index: 898

Text #8

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.2030. Min Act: -0.1700

Data Index: 1437250 (Python Code)

Max Activating Token Index: 680

Text #9

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1558. Min Act: -0.1700

Data Index: 1493712 (Python Code)

Max Activating Token Index: 550

Text #10

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1873. Min Act: -0.1700

Data Index: 725613 (C4 (Web Text))

Max Activating Token Index: 755

Text #11

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1715. Min Act: -0.1700

Data Index: 647461 (C4 (Web Text))

Max Activating Token Index: 770

Text #12

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1873. Min Act: -0.1700

Data Index: 547865 (C4 (Web Text))

Max Activating Token Index: 286

Text #13

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1873. Min Act: -0.1700

Data Index: 1593581 (Python Code)

Max Activating Token Index: 737

Text #14

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1715. Min Act: -0.1700

Data Index: 1458457 (Python Code)

Max Activating Token Index: 675

Text #15

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1715. Min Act: -0.1700

Data Index: 1578703 (Python Code)

Max Activating Token Index: 319

Text #16

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1873. Min Act: -0.1700

Data Index: 789127 (C4 (Web Text))

Max Activating Token Index: 890

Text #17

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1715. Min Act: -0.1700

Data Index: 941826 (C4 (Web Text))

Max Activating Token Index: 186

Text #18

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1715. Min Act: -0.1700

Data Index: 1552796 (Python Code)

Max Activating Token Index: 1023

Text #19

Max Range: 3.2659. Min Range: -3.2659

Max Act: 3.1558. Min Act: -0.1700

Data Index: 1443890 (Python Code)

Max Activating Token Index: 137