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

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

Neuron 1683 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 5.4506. Min Range: -5.4506

Max Act: 5.4506. Min Act: -0.1700

Data Index: 1490377 (Python Code)

Max Activating Token Index: 399

Text #1

Max Range: 5.4506. Min Range: -5.4506

Max Act: 5.1068. Min Act: -0.1700

Data Index: 1587805 (Python Code)

Max Activating Token Index: 444

Text #2

Max Range: 5.4506. Min Range: -5.4506

Max Act: 5.0443. Min Act: -0.1700

Data Index: 131736 (C4 (Web Text))

Max Activating Token Index: 693

Text #3

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.9506. Min Act: -0.1700

Data Index: 1073423 (C4 (Web Text))

Max Activating Token Index: 788

Text #4

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.9506. Min Act: -0.1700

Data Index: 1374177 (Python Code)

Max Activating Token Index: 564

Text #5

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.9193. Min Act: -0.1700

Data Index: 1464762 (Python Code)

Max Activating Token Index: 693

Text #6

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.9193. Min Act: -0.1700

Data Index: 246248 (C4 (Web Text))

Max Activating Token Index: 793

Text #7

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.9193. Min Act: -0.1700

Data Index: 1224017 (C4 (Web Text))

Max Activating Token Index: 680

Text #8

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.8881. Min Act: -0.1700

Data Index: 456150 (C4 (Web Text))

Max Activating Token Index: 632

Text #9

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.8256. Min Act: -0.1700

Data Index: 1348515 (C4 (Web Text))

Max Activating Token Index: 537

Text #10

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.8256. Min Act: -0.1700

Data Index: 291030 (C4 (Web Text))

Max Activating Token Index: 788

Text #11

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.7943. Min Act: -0.1700

Data Index: 998850 (C4 (Web Text))

Max Activating Token Index: 846

Text #12

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.7943. Min Act: -0.1700

Data Index: 955681 (C4 (Web Text))

Max Activating Token Index: 16

Text #13

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.8881. Min Act: -0.1700

Data Index: 1606694 (Python Code)

Max Activating Token Index: 120

Text #14

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.7943. Min Act: -0.1700

Data Index: 1582105 (Python Code)

Max Activating Token Index: 969

Text #15

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.8256. Min Act: -0.1700

Data Index: 1616782 (Python Code)

Max Activating Token Index: 445

Text #16

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.7943. Min Act: -0.1700

Data Index: 610527 (C4 (Web Text))

Max Activating Token Index: 697

Text #17

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.8256. Min Act: -0.1700

Data Index: 510688 (C4 (Web Text))

Max Activating Token Index: 807

Text #18

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.7943. Min Act: -0.1700

Data Index: 211195 (C4 (Web Text))

Max Activating Token Index: 694

Text #19

Max Range: 5.4506. Min Range: -5.4506

Max Act: 4.7631. Min Act: -0.1700

Data Index: 1333584 (C4 (Web Text))

Max Activating Token Index: 562