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

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

Neuron 1000 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.8860. Min Act: -0.1700

Data Index: 172164 (C4 (Web Text))

Max Activating Token Index: 921

Text #1

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.6983. Min Act: -0.1700

Data Index: 1317214 (C4 (Web Text))

Max Activating Token Index: 142

Text #2

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.6983. Min Act: -0.1700

Data Index: 229935 (C4 (Web Text))

Max Activating Token Index: 152

Text #3

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.6357. Min Act: -0.1700

Data Index: 944179 (C4 (Web Text))

Max Activating Token Index: 642

Text #4

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.6357. Min Act: -0.1700

Data Index: 1267347 (C4 (Web Text))

Max Activating Token Index: 706

Text #5

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.6670. Min Act: -0.1700

Data Index: 107807 (C4 (Web Text))

Max Activating Token Index: 632

Text #6

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5731. Min Act: -0.1700

Data Index: 1286261 (C4 (Web Text))

Max Activating Token Index: 727

Text #7

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5417. Min Act: -0.1700

Data Index: 501908 (C4 (Web Text))

Max Activating Token Index: 191

Text #8

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.6044. Min Act: -0.1700

Data Index: 1336344 (C4 (Web Text))

Max Activating Token Index: 797

Text #9

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.4791. Min Act: -0.1700

Data Index: 613259 (C4 (Web Text))

Max Activating Token Index: 861

Text #10

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5417. Min Act: -0.1700

Data Index: 582588 (C4 (Web Text))

Max Activating Token Index: 655

Text #11

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5417. Min Act: -0.1700

Data Index: 439572 (C4 (Web Text))

Max Activating Token Index: 203

Text #12

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5417. Min Act: -0.1700

Data Index: 967861 (C4 (Web Text))

Max Activating Token Index: 890

Text #13

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5731. Min Act: -0.1700

Data Index: 1233615 (C4 (Web Text))

Max Activating Token Index: 297

Text #14

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5104. Min Act: -0.1700

Data Index: 631773 (C4 (Web Text))

Max Activating Token Index: 89

Text #15

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5104. Min Act: -0.1700

Data Index: 714325 (C4 (Web Text))

Max Activating Token Index: 180

Text #16

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5104. Min Act: -0.1700

Data Index: 358531 (C4 (Web Text))

Max Activating Token Index: 780

Text #17

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5104. Min Act: -0.1700

Data Index: 249458 (C4 (Web Text))

Max Activating Token Index: 788

Text #18

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.5417. Min Act: -0.1700

Data Index: 1243121 (C4 (Web Text))

Max Activating Token Index: 658

Text #19

Max Range: 3.8860. Min Range: -3.8860

Max Act: 3.4791. Min Act: -0.1700

Data Index: 522936 (C4 (Web Text))

Max Activating Token Index: 814