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

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

Neuron 920 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.8804. Min Act: -0.1700

Data Index: 133433 (C4 (Web Text))

Max Activating Token Index: 577

Text #1

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.8491. Min Act: -0.1700

Data Index: 1643123 (Python Code)

Max Activating Token Index: 190

Text #2

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.5674. Min Act: -0.1700

Data Index: 388665 (C4 (Web Text))

Max Activating Token Index: 714

Text #3

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.5674. Min Act: -0.1700

Data Index: 2941 (C4 (Web Text))

Max Activating Token Index: 817

Text #4

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.5361. Min Act: -0.1700

Data Index: 989042 (C4 (Web Text))

Max Activating Token Index: 937

Text #5

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.5361. Min Act: -0.1700

Data Index: 1436544 (Python Code)

Max Activating Token Index: 799

Text #6

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.5204. Min Act: -0.1700

Data Index: 1556153 (Python Code)

Max Activating Token Index: 104

Text #7

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4891. Min Act: -0.1700

Data Index: 442454 (C4 (Web Text))

Max Activating Token Index: 582

Text #8

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4891. Min Act: -0.1700

Data Index: 1312580 (C4 (Web Text))

Max Activating Token Index: 656

Text #9

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4734. Min Act: -0.1700

Data Index: 860047 (C4 (Web Text))

Max Activating Token Index: 833

Text #10

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4734. Min Act: -0.1700

Data Index: 1598532 (Python Code)

Max Activating Token Index: 802

Text #11

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4891. Min Act: -0.1700

Data Index: 1039401 (C4 (Web Text))

Max Activating Token Index: 562

Text #12

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4891. Min Act: -0.1700

Data Index: 267777 (C4 (Web Text))

Max Activating Token Index: 687

Text #13

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4734. Min Act: -0.1700

Data Index: 132620 (C4 (Web Text))

Max Activating Token Index: 385

Text #14

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4578. Min Act: -0.1700

Data Index: 1518595 (Python Code)

Max Activating Token Index: 778

Text #15

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4578. Min Act: -0.1700

Data Index: 1042614 (C4 (Web Text))

Max Activating Token Index: 331

Text #16

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4578. Min Act: -0.1700

Data Index: 1223484 (C4 (Web Text))

Max Activating Token Index: 495

Text #17

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4734. Min Act: -0.1700

Data Index: 41168 (C4 (Web Text))

Max Activating Token Index: 877

Text #18

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4578. Min Act: -0.1700

Data Index: 1315140 (C4 (Web Text))

Max Activating Token Index: 947

Text #19

Max Range: 3.8804. Min Range: -3.8804

Max Act: 3.4421. Min Act: -0.1700

Data Index: 116683 (C4 (Web Text))

Max Activating Token Index: 387