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

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

Neuron 1972 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: 3.4424. Min Range: -3.4424

Max Act: 3.4424. Min Act: -0.1700

Data Index: 1493870 (Python Code)

Max Activating Token Index: 822

Text #1

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.3483. Min Act: -0.1700

Data Index: 1497096 (Python Code)

Max Activating Token Index: 702

Text #2

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2855. Min Act: -0.1700

Data Index: 950613 (C4 (Web Text))

Max Activating Token Index: 542

Text #3

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.3012. Min Act: -0.1700

Data Index: 649403 (C4 (Web Text))

Max Activating Token Index: 879

Text #4

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2855. Min Act: -0.1700

Data Index: 1627709 (Python Code)

Max Activating Token Index: 706

Text #5

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2540. Min Act: -0.1700

Data Index: 1199110 (C4 (Web Text))

Max Activating Token Index: 283

Text #6

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2383. Min Act: -0.1700

Data Index: 737674 (C4 (Web Text))

Max Activating Token Index: 794

Text #7

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2383. Min Act: -0.1700

Data Index: 1400292 (Python Code)

Max Activating Token Index: 865

Text #8

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2383. Min Act: -0.1700

Data Index: 1369435 (Python Code)

Max Activating Token Index: 881

Text #9

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2383. Min Act: -0.1700

Data Index: 265476 (C4 (Web Text))

Max Activating Token Index: 858

Text #10

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2226. Min Act: -0.1700

Data Index: 658548 (C4 (Web Text))

Max Activating Token Index: 700

Text #11

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2226. Min Act: -0.1700

Data Index: 1360769 (Python Code)

Max Activating Token Index: 595

Text #12

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.1911. Min Act: -0.1700

Data Index: 412849 (C4 (Web Text))

Max Activating Token Index: 875

Text #13

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.2226. Min Act: -0.1700

Data Index: 1006915 (C4 (Web Text))

Max Activating Token Index: 40

Text #14

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.1911. Min Act: -0.1700

Data Index: 1064900 (C4 (Web Text))

Max Activating Token Index: 222

Text #15

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.1911. Min Act: -0.1700

Data Index: 1633208 (Python Code)

Max Activating Token Index: 192

Text #16

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.1754. Min Act: -0.1700

Data Index: 1439836 (Python Code)

Max Activating Token Index: 305

Text #17

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.1911. Min Act: -0.1700

Data Index: 1633774 (Python Code)

Max Activating Token Index: 481

Text #18

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.1911. Min Act: -0.1700

Data Index: 971813 (C4 (Web Text))

Max Activating Token Index: 576

Text #19

Max Range: 3.4424. Min Range: -3.4424

Max Act: 3.1596. Min Act: -0.1700

Data Index: 349472 (C4 (Web Text))

Max Activating Token Index: 267