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

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

Neuron 165 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: 5.8352. Min Range: -5.8352

Max Act: 5.8352. Min Act: -0.1700

Data Index: 1398193 (Python Code)

Max Activating Token Index: 614

Text #1

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.6164. Min Act: -0.1700

Data Index: 379423 (C4 (Web Text))

Max Activating Token Index: 869

Text #2

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.4602. Min Act: -0.1700

Data Index: 1131950 (C4 (Web Text))

Max Activating Token Index: 827

Text #3

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.4914. Min Act: -0.1700

Data Index: 954562 (C4 (Web Text))

Max Activating Token Index: 738

Text #4

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.4914. Min Act: -0.1700

Data Index: 740412 (C4 (Web Text))

Max Activating Token Index: 659

Text #5

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.4602. Min Act: -0.1700

Data Index: 635247 (C4 (Web Text))

Max Activating Token Index: 390

Text #6

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.4289. Min Act: -0.1700

Data Index: 452764 (C4 (Web Text))

Max Activating Token Index: 252

Text #7

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.3977. Min Act: -0.1700

Data Index: 1286193 (C4 (Web Text))

Max Activating Token Index: 842

Text #8

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.3977. Min Act: -0.1700

Data Index: 221873 (C4 (Web Text))

Max Activating Token Index: 71

Text #9

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.3039. Min Act: -0.1700

Data Index: 1640133 (Python Code)

Max Activating Token Index: 341

Text #10

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.2102. Min Act: -0.1700

Data Index: 38084 (C4 (Web Text))

Max Activating Token Index: 160

Text #11

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.2727. Min Act: -0.1700

Data Index: 1160187 (C4 (Web Text))

Max Activating Token Index: 995

Text #12

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.1789. Min Act: -0.1700

Data Index: 413858 (C4 (Web Text))

Max Activating Token Index: 741

Text #13

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.0539. Min Act: -0.1700

Data Index: 475330 (C4 (Web Text))

Max Activating Token Index: 733

Text #14

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.1789. Min Act: -0.1700

Data Index: 1397438 (Python Code)

Max Activating Token Index: 550

Text #15

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.1477. Min Act: -0.1700

Data Index: 868024 (C4 (Web Text))

Max Activating Token Index: 503

Text #16

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.0852. Min Act: -0.1700

Data Index: 721165 (C4 (Web Text))

Max Activating Token Index: 790

Text #17

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.0852. Min Act: -0.1700

Data Index: 1041743 (C4 (Web Text))

Max Activating Token Index: 816

Text #18

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.0852. Min Act: -0.1700

Data Index: 345207 (C4 (Web Text))

Max Activating Token Index: 538

Text #19

Max Range: 5.8352. Min Range: -5.8352

Max Act: 5.0227. Min Act: -0.1700

Data Index: 907260 (C4 (Web Text))

Max Activating Token Index: 115