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

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

Neuron 172 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')

Text #0

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.7091. Min Act: -0.1700

Data Index: 506256 (C4 (Web Text))

Max Activating Token Index: 490

Text #1

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.6935. Min Act: -0.1700

Data Index: 484986 (C4 (Web Text))

Max Activating Token Index: 448

Text #2

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.6622. Min Act: -0.1700

Data Index: 1601415 (Python Code)

Max Activating Token Index: 270

Text #3

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.6465. Min Act: -0.1700

Data Index: 938076 (C4 (Web Text))

Max Activating Token Index: 461

Text #4

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5839. Min Act: -0.1700

Data Index: 546384 (C4 (Web Text))

Max Activating Token Index: 404

Text #5

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.6152. Min Act: -0.1700

Data Index: 1482006 (Python Code)

Max Activating Token Index: 877

Text #6

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.6152. Min Act: -0.1700

Data Index: 372800 (C4 (Web Text))

Max Activating Token Index: 429

Text #7

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5682. Min Act: -0.1700

Data Index: 1644372 (Python Code)

Max Activating Token Index: 1003

Text #8

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5526. Min Act: -0.1700

Data Index: 1653121 (Python Code)

Max Activating Token Index: 1013

Text #9

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.6778. Min Act: -0.1699

Data Index: 1607851 (Python Code)

Max Activating Token Index: 1002

Text #10

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5526. Min Act: -0.1700

Data Index: 1404856 (Python Code)

Max Activating Token Index: 795

Text #11

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5995. Min Act: -0.1700

Data Index: 1509220 (Python Code)

Max Activating Token Index: 1014

Text #12

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5526. Min Act: -0.1700

Data Index: 1625988 (Python Code)

Max Activating Token Index: 674

Text #13

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5995. Min Act: -0.1700

Data Index: 1494651 (Python Code)

Max Activating Token Index: 893

Text #14

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5212. Min Act: -0.1699

Data Index: 1385314 (Python Code)

Max Activating Token Index: 1020

Text #15

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5995. Min Act: -0.1700

Data Index: 683115 (C4 (Web Text))

Max Activating Token Index: 895

Text #16

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5995. Min Act: -0.1700

Data Index: 1322698 (C4 (Web Text))

Max Activating Token Index: 785

Text #17

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5839. Min Act: -0.1700

Data Index: 93958 (C4 (Web Text))

Max Activating Token Index: 349

Text #18

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5995. Min Act: -0.1700

Data Index: 432688 (C4 (Web Text))

Max Activating Token Index: 368

Text #19

Max Range: 3.7091. Min Range: -3.7091

Max Act: 3.5682. Min Act: -0.1698

Data Index: 1437396 (Python Code)

Max Activating Token Index: 610