Model: GELU Model: 2 Layers, 2048 Neurons per Layer

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

Neuron 349 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-2l')

Text #0

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.8882. Min Act: -0.1700

Data Index: 823237 (C4 (Web Text))

Max Activating Token Index: 88

Text #1

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.8257. Min Act: -0.1700

Data Index: 1168549 (C4 (Web Text))

Max Activating Token Index: 790

Text #2

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7788. Min Act: -0.1700

Data Index: 1619487 (Python Code)

Max Activating Token Index: 376

Text #3

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7944. Min Act: -0.1700

Data Index: 234088 (C4 (Web Text))

Max Activating Token Index: 444

Text #4

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7788. Min Act: -0.1700

Data Index: 950525 (C4 (Web Text))

Max Activating Token Index: 837

Text #5

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7318. Min Act: -0.1700

Data Index: 1339252 (C4 (Web Text))

Max Activating Token Index: 534

Text #6

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7318. Min Act: -0.1700

Data Index: 1569532 (Python Code)

Max Activating Token Index: 733

Text #7

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7475. Min Act: -0.1700

Data Index: 443802 (C4 (Web Text))

Max Activating Token Index: 399

Text #8

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7162. Min Act: -0.1700

Data Index: 840268 (C4 (Web Text))

Max Activating Token Index: 861

Text #9

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7162. Min Act: -0.1700

Data Index: 269458 (C4 (Web Text))

Max Activating Token Index: 648

Text #10

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7005. Min Act: -0.1699

Data Index: 1592887 (Python Code)

Max Activating Token Index: 184

Text #11

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.6849. Min Act: -0.1700

Data Index: 570104 (C4 (Web Text))

Max Activating Token Index: 891

Text #12

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7005. Min Act: -0.1700

Data Index: 1102652 (C4 (Web Text))

Max Activating Token Index: 153

Text #13

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7005. Min Act: -0.1700

Data Index: 622629 (C4 (Web Text))

Max Activating Token Index: 365

Text #14

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.6536. Min Act: -0.1700

Data Index: 1535672 (Python Code)

Max Activating Token Index: 973

Text #15

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.7005. Min Act: -0.1700

Data Index: 490744 (C4 (Web Text))

Max Activating Token Index: 458

Text #16

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.6692. Min Act: -0.1700

Data Index: 31960 (C4 (Web Text))

Max Activating Token Index: 850

Text #17

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.6849. Min Act: -0.1700

Data Index: 572810 (C4 (Web Text))

Max Activating Token Index: 889

Text #18

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.6692. Min Act: -0.1700

Data Index: 982814 (C4 (Web Text))

Max Activating Token Index: 310

Text #19

Max Range: 3.8882. Min Range: -3.8882

Max Act: 3.6536. Min Act: -0.1700

Data Index: 1498174 (Python Code)

Max Activating Token Index: 809