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

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

Neuron 149 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.6662. Min Act: -0.1700

Data Index: 977850 (C4 (Web Text))

Max Activating Token Index: 722

Text #1

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.6349. Min Act: -0.1700

Data Index: 96507 (C4 (Web Text))

Max Activating Token Index: 731

Text #2

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5723. Min Act: -0.1700

Data Index: 632008 (C4 (Web Text))

Max Activating Token Index: 656

Text #3

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5879. Min Act: -0.1700

Data Index: 44198 (C4 (Web Text))

Max Activating Token Index: 631

Text #4

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5879. Min Act: -0.1700

Data Index: 919411 (C4 (Web Text))

Max Activating Token Index: 555

Text #5

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5566. Min Act: -0.1700

Data Index: 1486972 (Python Code)

Max Activating Token Index: 745

Text #6

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5566. Min Act: -0.1700

Data Index: 532472 (C4 (Web Text))

Max Activating Token Index: 122

Text #7

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5253. Min Act: -0.1700

Data Index: 492023 (C4 (Web Text))

Max Activating Token Index: 755

Text #8

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.4939. Min Act: -0.1700

Data Index: 1079449 (C4 (Web Text))

Max Activating Token Index: 647

Text #9

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5096. Min Act: -0.1700

Data Index: 1565015 (Python Code)

Max Activating Token Index: 640

Text #10

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5096. Min Act: -0.1700

Data Index: 1214733 (C4 (Web Text))

Max Activating Token Index: 730

Text #11

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5253. Min Act: -0.1700

Data Index: 520519 (C4 (Web Text))

Max Activating Token Index: 398

Text #12

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5096. Min Act: -0.1700

Data Index: 1445951 (Python Code)

Max Activating Token Index: 929

Text #13

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.4939. Min Act: -0.1700

Data Index: 1234946 (C4 (Web Text))

Max Activating Token Index: 648

Text #14

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5096. Min Act: -0.1700

Data Index: 986152 (C4 (Web Text))

Max Activating Token Index: 870

Text #15

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.4783. Min Act: -0.1700

Data Index: 848020 (C4 (Web Text))

Max Activating Token Index: 406

Text #16

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.4939. Min Act: -0.1700

Data Index: 42810 (C4 (Web Text))

Max Activating Token Index: 546

Text #17

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.4783. Min Act: -0.1700

Data Index: 411092 (C4 (Web Text))

Max Activating Token Index: 639

Text #18

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.4783. Min Act: -0.1700

Data Index: 746205 (C4 (Web Text))

Max Activating Token Index: 815

Text #19

Max Range: 3.6662. Min Range: -3.6662

Max Act: 3.5096. Min Act: -0.1700

Data Index: 860891 (C4 (Web Text))

Max Activating Token Index: 511