Model: GELU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 2038 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 5.2581. Min Range: -5.2581

Max Act: 5.2581. Min Act: -0.1700

Data Index: 1576708 (Python Code)

Max Activating Token Index: 950

Text #1

Max Range: 5.2581. Min Range: -5.2581

Max Act: 5.1019. Min Act: -0.1700

Data Index: 803023 (C4 (Web Text))

Max Activating Token Index: 703

Text #2

Max Range: 5.2581. Min Range: -5.2581

Max Act: 5.0706. Min Act: -0.1700

Data Index: 306727 (C4 (Web Text))

Max Activating Token Index: 455

Text #3

Max Range: 5.2581. Min Range: -5.2581

Max Act: 5.0081. Min Act: -0.1700

Data Index: 1427041 (Python Code)

Max Activating Token Index: 37

Text #4

Max Range: 5.2581. Min Range: -5.2581

Max Act: 5.0394. Min Act: -0.1700

Data Index: 682393 (C4 (Web Text))

Max Activating Token Index: 16

Text #5

Max Range: 5.2581. Min Range: -5.2581

Max Act: 5.0081. Min Act: -0.1700

Data Index: 363675 (C4 (Web Text))

Max Activating Token Index: 196

Text #6

Max Range: 5.2581. Min Range: -5.2581

Max Act: 5.0081. Min Act: -0.1700

Data Index: 472857 (C4 (Web Text))

Max Activating Token Index: 198

Text #7

Max Range: 5.2581. Min Range: -5.2581

Max Act: 5.0081. Min Act: -0.1700

Data Index: 1629853 (Python Code)

Max Activating Token Index: 702

Text #8

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9456. Min Act: -0.1700

Data Index: 989527 (C4 (Web Text))

Max Activating Token Index: 59

Text #9

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9456. Min Act: -0.1700

Data Index: 268278 (C4 (Web Text))

Max Activating Token Index: 351

Text #10

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9769. Min Act: -0.1700

Data Index: 620346 (C4 (Web Text))

Max Activating Token Index: 151

Text #11

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9769. Min Act: -0.1700

Data Index: 449636 (C4 (Web Text))

Max Activating Token Index: 370

Text #12

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9144. Min Act: -0.1700

Data Index: 694785 (C4 (Web Text))

Max Activating Token Index: 463

Text #13

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9456. Min Act: -0.1700

Data Index: 138438 (C4 (Web Text))

Max Activating Token Index: 415

Text #14

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9769. Min Act: -0.1700

Data Index: 1552536 (Python Code)

Max Activating Token Index: 311

Text #15

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.8831. Min Act: -0.1700

Data Index: 1647724 (Python Code)

Max Activating Token Index: 516

Text #16

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9769. Min Act: -0.1700

Data Index: 1528141 (Python Code)

Max Activating Token Index: 997

Text #17

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9456. Min Act: -0.1700

Data Index: 1367191 (Python Code)

Max Activating Token Index: 414

Text #18

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9769. Min Act: -0.1700

Data Index: 403141 (C4 (Web Text))

Max Activating Token Index: 209

Text #19

Max Range: 5.2581. Min Range: -5.2581

Max Act: 4.9456. Min Act: -0.1700

Data Index: 672098 (C4 (Web Text))

Max Activating Token Index: 657