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

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

Neuron 870 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: 5.1856. Min Range: -5.1856

Max Act: 5.1856. Min Act: -0.1700

Data Index: 1121691 (C4 (Web Text))

Max Activating Token Index: 698

Text #1

Max Range: 5.1856. Min Range: -5.1856

Max Act: 5.0450. Min Act: -0.1700

Data Index: 861323 (C4 (Web Text))

Max Activating Token Index: 637

Text #2

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.9668. Min Act: -0.1700

Data Index: 1115133 (C4 (Web Text))

Max Activating Token Index: 104

Text #3

Max Range: 5.1856. Min Range: -5.1856

Max Act: 5.0450. Min Act: -0.1700

Data Index: 667292 (C4 (Web Text))

Max Activating Token Index: 658

Text #4

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.9825. Min Act: -0.1699

Data Index: 1132091 (C4 (Web Text))

Max Activating Token Index: 77

Text #5

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.9825. Min Act: -0.1700

Data Index: 290657 (C4 (Web Text))

Max Activating Token Index: 280

Text #6

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.9043. Min Act: -0.1700

Data Index: 99122 (C4 (Web Text))

Max Activating Token Index: 638

Text #7

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8731. Min Act: -0.1664

Data Index: 1359408 (C4 (Web Text))

Max Activating Token Index: 74

Text #8

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.9043. Min Act: -0.1700

Data Index: 1050591 (C4 (Web Text))

Max Activating Token Index: 908

Text #9

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8418. Min Act: -0.1700

Data Index: 789970 (C4 (Web Text))

Max Activating Token Index: 592

Text #10

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.9043. Min Act: -0.1700

Data Index: 1172473 (C4 (Web Text))

Max Activating Token Index: 228

Text #11

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8887. Min Act: -0.1700

Data Index: 131894 (C4 (Web Text))

Max Activating Token Index: 49

Text #12

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8106. Min Act: -0.1700

Data Index: 1040273 (C4 (Web Text))

Max Activating Token Index: 91

Text #13

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8731. Min Act: -0.1700

Data Index: 460325 (C4 (Web Text))

Max Activating Token Index: 637

Text #14

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8731. Min Act: -0.1700

Data Index: 602546 (C4 (Web Text))

Max Activating Token Index: 804

Text #15

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8106. Min Act: -0.1700

Data Index: 980342 (C4 (Web Text))

Max Activating Token Index: 554

Text #16

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8575. Min Act: -0.1698

Data Index: 1149932 (C4 (Web Text))

Max Activating Token Index: 85

Text #17

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8418. Min Act: -0.1700

Data Index: 1141332 (C4 (Web Text))

Max Activating Token Index: 365

Text #18

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.8731. Min Act: -0.1700

Data Index: 745542 (C4 (Web Text))

Max Activating Token Index: 45

Text #19

Max Range: 5.1856. Min Range: -5.1856

Max Act: 4.9200. Min Act: -0.1700

Data Index: 333888 (C4 (Web Text))

Max Activating Token Index: 737