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

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

Neuron 1542 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: 5.4230. Min Range: -5.4230

Max Act: 5.4230. Min Act: -0.1700

Data Index: 97833 (C4 (Web Text))

Max Activating Token Index: 89

Text #1

Max Range: 5.4230. Min Range: -5.4230

Max Act: 5.1886. Min Act: -0.1700

Data Index: 732343 (C4 (Web Text))

Max Activating Token Index: 405

Text #2

Max Range: 5.4230. Min Range: -5.4230

Max Act: 5.0324. Min Act: -0.1700

Data Index: 1316575 (C4 (Web Text))

Max Activating Token Index: 122

Text #3

Max Range: 5.4230. Min Range: -5.4230

Max Act: 5.0011. Min Act: -0.1700

Data Index: 1003441 (C4 (Web Text))

Max Activating Token Index: 751

Text #4

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.9699. Min Act: -0.1700

Data Index: 70041 (C4 (Web Text))

Max Activating Token Index: 441

Text #5

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.9074. Min Act: -0.1700

Data Index: 224509 (C4 (Web Text))

Max Activating Token Index: 850

Text #6

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.9230. Min Act: -0.1700

Data Index: 95178 (C4 (Web Text))

Max Activating Token Index: 267

Text #7

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.9543. Min Act: -0.1700

Data Index: 862513 (C4 (Web Text))

Max Activating Token Index: 820

Text #8

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.8918. Min Act: -0.1700

Data Index: 40134 (C4 (Web Text))

Max Activating Token Index: 241

Text #9

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.8605. Min Act: -0.1700

Data Index: 347048 (C4 (Web Text))

Max Activating Token Index: 258

Text #10

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.8449. Min Act: -0.1700

Data Index: 667964 (C4 (Web Text))

Max Activating Token Index: 60

Text #11

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.7511. Min Act: -0.1700

Data Index: 10635 (C4 (Web Text))

Max Activating Token Index: 926

Text #12

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.6730. Min Act: -0.1700

Data Index: 335386 (C4 (Web Text))

Max Activating Token Index: 44

Text #13

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.6886. Min Act: -0.1700

Data Index: 402236 (C4 (Web Text))

Max Activating Token Index: 66

Text #14

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.6574. Min Act: -0.1700

Data Index: 904117 (C4 (Web Text))

Max Activating Token Index: 148

Text #15

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.7355. Min Act: -0.1700

Data Index: 418303 (C4 (Web Text))

Max Activating Token Index: 803

Text #16

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.6886. Min Act: -0.1700

Data Index: 36189 (C4 (Web Text))

Max Activating Token Index: 885

Text #17

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.5324. Min Act: -0.1700

Data Index: 27647 (C4 (Web Text))

Max Activating Token Index: 751

Text #18

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.6418. Min Act: -0.1700

Data Index: 1065333 (C4 (Web Text))

Max Activating Token Index: 78

Text #19

Max Range: 5.4230. Min Range: -5.4230

Max Act: 4.6105. Min Act: -0.1700

Data Index: 636380 (C4 (Web Text))

Max Activating Token Index: 572