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

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

Neuron 402 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 4.0437. Min Range: -4.0437

Max Act: 4.0437. Min Act: -0.1700

Data Index: 749334 (C4 (Web Text))

Max Activating Token Index: 177

Text #1

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.9499. Min Act: -0.1700

Data Index: 1380253 (Python Code)

Max Activating Token Index: 942

Text #2

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.8678. Min Act: -0.1700

Data Index: 400386 (C4 (Web Text))

Max Activating Token Index: 568

Text #3

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.9186. Min Act: -0.1700

Data Index: 539646 (C4 (Web Text))

Max Activating Token Index: 1001

Text #4

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.9186. Min Act: -0.1700

Data Index: 927018 (C4 (Web Text))

Max Activating Token Index: 264

Text #5

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.8912. Min Act: -0.1700

Data Index: 1522577 (Python Code)

Max Activating Token Index: 498

Text #6

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.8209. Min Act: -0.1700

Data Index: 1496492 (Python Code)

Max Activating Token Index: 545

Text #7

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.8287. Min Act: -0.1700

Data Index: 1487132 (Python Code)

Max Activating Token Index: 774

Text #8

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.8209. Min Act: -0.1700

Data Index: 447774 (C4 (Web Text))

Max Activating Token Index: 37

Text #9

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.7974. Min Act: -0.1700

Data Index: 1489209 (Python Code)

Max Activating Token Index: 300

Text #10

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.7778. Min Act: -0.1700

Data Index: 590082 (C4 (Web Text))

Max Activating Token Index: 782

Text #11

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.8209. Min Act: -0.1700

Data Index: 1260508 (C4 (Web Text))

Max Activating Token Index: 629

Text #12

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.7153. Min Act: -0.1700

Data Index: 921910 (C4 (Web Text))

Max Activating Token Index: 473

Text #13

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.8130. Min Act: -0.1700

Data Index: 170571 (C4 (Web Text))

Max Activating Token Index: 755

Text #14

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.6918. Min Act: -0.1700

Data Index: 1352145 (C4 (Web Text))

Max Activating Token Index: 744

Text #15

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.7778. Min Act: -0.1700

Data Index: 1624692 (Python Code)

Max Activating Token Index: 917

Text #16

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.7309. Min Act: -0.1700

Data Index: 1565077 (Python Code)

Max Activating Token Index: 694

Text #17

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.7231. Min Act: -0.1700

Data Index: 1309898 (C4 (Web Text))

Max Activating Token Index: 637

Text #18

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.6722. Min Act: -0.1700

Data Index: 9446 (C4 (Web Text))

Max Activating Token Index: 588

Text #19

Max Range: 4.0437. Min Range: -4.0437

Max Act: 3.6840. Min Act: -0.1700

Data Index: 878589 (C4 (Web Text))

Max Activating Token Index: 758