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

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

Neuron 1063 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.2359. Min Act: -0.1700

Data Index: 831557 (C4 (Web Text))

Max Activating Token Index: 971

Text #1

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.2202. Min Act: -0.1700

Data Index: 406061 (C4 (Web Text))

Max Activating Token Index: 1019

Text #2

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.2359. Min Act: -0.1700

Data Index: 123616 (C4 (Web Text))

Max Activating Token Index: 921

Text #3

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1415. Min Act: -0.1699

Data Index: 1468292 (Python Code)

Max Activating Token Index: 379

Text #4

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1572. Min Act: -0.1700

Data Index: 1056142 (C4 (Web Text))

Max Activating Token Index: 956

Text #5

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1730. Min Act: -0.1700

Data Index: 632900 (C4 (Web Text))

Max Activating Token Index: 816

Text #6

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1730. Min Act: -0.1700

Data Index: 547955 (C4 (Web Text))

Max Activating Token Index: 724

Text #7

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1257. Min Act: -0.1700

Data Index: 1536300 (Python Code)

Max Activating Token Index: 830

Text #8

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1415. Min Act: -0.1700

Data Index: 1582158 (Python Code)

Max Activating Token Index: 744

Text #9

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1572. Min Act: -0.1700

Data Index: 1133659 (C4 (Web Text))

Max Activating Token Index: 547

Text #10

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1415. Min Act: -0.1700

Data Index: 484304 (C4 (Web Text))

Max Activating Token Index: 934

Text #11

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1415. Min Act: -0.1700

Data Index: 59522 (C4 (Web Text))

Max Activating Token Index: 673

Text #12

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1100. Min Act: -0.1700

Data Index: 347140 (C4 (Web Text))

Max Activating Token Index: 391

Text #13

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1257. Min Act: -0.1700

Data Index: 1113740 (C4 (Web Text))

Max Activating Token Index: 476

Text #14

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1100. Min Act: -0.1700

Data Index: 555195 (C4 (Web Text))

Max Activating Token Index: 736

Text #15

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1100. Min Act: -0.1700

Data Index: 30496 (C4 (Web Text))

Max Activating Token Index: 783

Text #16

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1100. Min Act: -0.1700

Data Index: 1432160 (Python Code)

Max Activating Token Index: 882

Text #17

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.0942. Min Act: -0.1700

Data Index: 138684 (C4 (Web Text))

Max Activating Token Index: 808

Text #18

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1100. Min Act: -0.1700

Data Index: 371963 (C4 (Web Text))

Max Activating Token Index: 535

Text #19

Max Range: 3.2359. Min Range: -3.2359

Max Act: 3.1100. Min Act: -0.1700

Data Index: 540458 (C4 (Web Text))

Max Activating Token Index: 750