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

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

Neuron 1067 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: 5.1554. Min Range: -5.1554

Max Act: 5.1554. Min Act: -0.1700

Data Index: 1055477 (C4 (Web Text))

Max Activating Token Index: 448

Text #1

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.8741. Min Act: -0.1700

Data Index: 1106345 (C4 (Web Text))

Max Activating Token Index: 93

Text #2

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.7803. Min Act: -0.1700

Data Index: 1032371 (C4 (Web Text))

Max Activating Token Index: 675

Text #3

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.7491. Min Act: -0.1700

Data Index: 726368 (C4 (Web Text))

Max Activating Token Index: 955

Text #4

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.6866. Min Act: -0.1699

Data Index: 363893 (C4 (Web Text))

Max Activating Token Index: 592

Text #5

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.6866. Min Act: -0.1700

Data Index: 860063 (C4 (Web Text))

Max Activating Token Index: 237

Text #6

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.6866. Min Act: -0.1700

Data Index: 1181271 (C4 (Web Text))

Max Activating Token Index: 695

Text #7

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.6866. Min Act: -0.1700

Data Index: 793876 (C4 (Web Text))

Max Activating Token Index: 195

Text #8

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.6241. Min Act: -0.1700

Data Index: 1233153 (C4 (Web Text))

Max Activating Token Index: 513

Text #9

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.5928. Min Act: -0.1700

Data Index: 239271 (C4 (Web Text))

Max Activating Token Index: 158

Text #10

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.6241. Min Act: -0.1700

Data Index: 158191 (C4 (Web Text))

Max Activating Token Index: 150

Text #11

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.5616. Min Act: -0.1699

Data Index: 1097946 (C4 (Web Text))

Max Activating Token Index: 97

Text #12

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.6553. Min Act: -0.1700

Data Index: 162948 (C4 (Web Text))

Max Activating Token Index: 218

Text #13

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.5928. Min Act: -0.1700

Data Index: 659420 (C4 (Web Text))

Max Activating Token Index: 396

Text #14

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.5928. Min Act: -0.1700

Data Index: 1635061 (Python Code)

Max Activating Token Index: 612

Text #15

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.5303. Min Act: -0.1700

Data Index: 295454 (C4 (Web Text))

Max Activating Token Index: 487

Text #16

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.5303. Min Act: -0.1700

Data Index: 795938 (C4 (Web Text))

Max Activating Token Index: 72

Text #17

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.5616. Min Act: -0.1700

Data Index: 814907 (C4 (Web Text))

Max Activating Token Index: 581

Text #18

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.5616. Min Act: -0.1700

Data Index: 919616 (C4 (Web Text))

Max Activating Token Index: 394

Text #19

Max Range: 5.1554. Min Range: -5.1554

Max Act: 4.5303. Min Act: -0.1699

Data Index: 373676 (C4 (Web Text))

Max Activating Token Index: 444