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

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

Neuron 77 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: 5.1497. Min Range: -5.1497

Max Act: 5.1497. Min Act: -0.1700

Data Index: 13505 (C4 (Web Text))

Max Activating Token Index: 721

Text #1

Max Range: 5.1497. Min Range: -5.1497

Max Act: 5.0872. Min Act: -0.1700

Data Index: 224807 (C4 (Web Text))

Max Activating Token Index: 114

Text #2

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.9310. Min Act: -0.1700

Data Index: 1532464 (Python Code)

Max Activating Token Index: 600

Text #3

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.9310. Min Act: -0.1700

Data Index: 867102 (C4 (Web Text))

Max Activating Token Index: 673

Text #4

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.9310. Min Act: -0.1700

Data Index: 334031 (C4 (Web Text))

Max Activating Token Index: 264

Text #5

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.9310. Min Act: -0.1700

Data Index: 642262 (C4 (Web Text))

Max Activating Token Index: 280

Text #6

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8997. Min Act: -0.1700

Data Index: 628956 (C4 (Web Text))

Max Activating Token Index: 295

Text #7

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8997. Min Act: -0.1700

Data Index: 262745 (C4 (Web Text))

Max Activating Token Index: 86

Text #8

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8997. Min Act: -0.1700

Data Index: 1584780 (Python Code)

Max Activating Token Index: 45

Text #9

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8685. Min Act: -0.1700

Data Index: 1516153 (Python Code)

Max Activating Token Index: 573

Text #10

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8372. Min Act: -0.1700

Data Index: 1415571 (Python Code)

Max Activating Token Index: 49

Text #11

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8372. Min Act: -0.1700

Data Index: 227826 (C4 (Web Text))

Max Activating Token Index: 924

Text #12

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8685. Min Act: -0.1700

Data Index: 1074125 (C4 (Web Text))

Max Activating Token Index: 22

Text #13

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8372. Min Act: -0.1700

Data Index: 888025 (C4 (Web Text))

Max Activating Token Index: 158

Text #14

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.9310. Min Act: -0.1700

Data Index: 381037 (C4 (Web Text))

Max Activating Token Index: 788

Text #15

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8372. Min Act: -0.1700

Data Index: 987308 (C4 (Web Text))

Max Activating Token Index: 801

Text #16

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8060. Min Act: -0.1700

Data Index: 1466458 (Python Code)

Max Activating Token Index: 52

Text #17

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8060. Min Act: -0.1700

Data Index: 1375851 (Python Code)

Max Activating Token Index: 237

Text #18

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8372. Min Act: -0.1700

Data Index: 1650532 (Python Code)

Max Activating Token Index: 47

Text #19

Max Range: 5.1497. Min Range: -5.1497

Max Act: 4.8372. Min Act: -0.1700

Data Index: 1406172 (Python Code)

Max Activating Token Index: 51