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

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

Neuron 643 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: 3.3517. Min Range: -3.3517

Max Act: 3.3517. Min Act: -0.1700

Data Index: 470172 (C4 (Web Text))

Max Activating Token Index: 283

Text #1

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.2575. Min Act: -0.1700

Data Index: 1010172 (C4 (Web Text))

Max Activating Token Index: 823

Text #2

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.2575. Min Act: -0.1700

Data Index: 112436 (C4 (Web Text))

Max Activating Token Index: 981

Text #3

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.2417. Min Act: -0.1700

Data Index: 1553890 (Python Code)

Max Activating Token Index: 1016

Text #4

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.2260. Min Act: -0.1700

Data Index: 350393 (C4 (Web Text))

Max Activating Token Index: 889

Text #5

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1946. Min Act: -0.1700

Data Index: 1118194 (C4 (Web Text))

Max Activating Token Index: 747

Text #6

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1946. Min Act: -0.1700

Data Index: 264491 (C4 (Web Text))

Max Activating Token Index: 198

Text #7

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.2260. Min Act: -0.1700

Data Index: 1379233 (Python Code)

Max Activating Token Index: 41

Text #8

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1946. Min Act: -0.1700

Data Index: 1498486 (Python Code)

Max Activating Token Index: 297

Text #9

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1631. Min Act: -0.1700

Data Index: 1528147 (Python Code)

Max Activating Token Index: 399

Text #10

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.2103. Min Act: -0.1700

Data Index: 187839 (C4 (Web Text))

Max Activating Token Index: 822

Text #11

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1631. Min Act: -0.1700

Data Index: 56915 (C4 (Web Text))

Max Activating Token Index: 809

Text #12

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1473. Min Act: -0.1700

Data Index: 103681 (C4 (Web Text))

Max Activating Token Index: 631

Text #13

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1788. Min Act: -0.1700

Data Index: 1640092 (Python Code)

Max Activating Token Index: 895

Text #14

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1946. Min Act: -0.1700

Data Index: 728876 (C4 (Web Text))

Max Activating Token Index: 378

Text #15

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1316. Min Act: -0.1700

Data Index: 10318 (C4 (Web Text))

Max Activating Token Index: 346

Text #16

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1631. Min Act: -0.1700

Data Index: 1611636 (Python Code)

Max Activating Token Index: 650

Text #17

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1316. Min Act: -0.1700

Data Index: 1004273 (C4 (Web Text))

Max Activating Token Index: 928

Text #18

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1316. Min Act: -0.1700

Data Index: 173788 (C4 (Web Text))

Max Activating Token Index: 642

Text #19

Max Range: 3.3517. Min Range: -3.3517

Max Act: 3.1473. Min Act: -0.1700

Data Index: 743728 (C4 (Web Text))

Max Activating Token Index: 529