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

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

Neuron 1760 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: 5.7919. Min Range: -5.7919

Max Act: 5.7607. Min Act: -0.1700

Data Index: 291168 (C4 (Web Text))

Max Activating Token Index: 569

Text #1

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.7919. Min Act: -0.1700

Data Index: 116520 (C4 (Web Text))

Max Activating Token Index: 864

Text #2

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.7607. Min Act: -0.1700

Data Index: 543476 (C4 (Web Text))

Max Activating Token Index: 862

Text #3

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.7294. Min Act: -0.1700

Data Index: 1463816 (Python Code)

Max Activating Token Index: 517

Text #4

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.7607. Min Act: -0.1700

Data Index: 1146423 (C4 (Web Text))

Max Activating Token Index: 492

Text #5

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6982. Min Act: -0.1700

Data Index: 450400 (C4 (Web Text))

Max Activating Token Index: 569

Text #6

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.7607. Min Act: -0.1700

Data Index: 1606032 (Python Code)

Max Activating Token Index: 614

Text #7

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6357. Min Act: -0.1700

Data Index: 1608985 (Python Code)

Max Activating Token Index: 286

Text #8

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6669. Min Act: -0.1700

Data Index: 1638916 (Python Code)

Max Activating Token Index: 218

Text #9

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6982. Min Act: -0.1700

Data Index: 1059664 (C4 (Web Text))

Max Activating Token Index: 488

Text #10

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6669. Min Act: -0.1700

Data Index: 1378150 (Python Code)

Max Activating Token Index: 484

Text #11

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6357. Min Act: -0.1700

Data Index: 1561348 (Python Code)

Max Activating Token Index: 1012

Text #12

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6357. Min Act: -0.1700

Data Index: 1415056 (Python Code)

Max Activating Token Index: 615

Text #13

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6982. Min Act: -0.1700

Data Index: 1567820 (Python Code)

Max Activating Token Index: 584

Text #14

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6669. Min Act: -0.1700

Data Index: 981779 (C4 (Web Text))

Max Activating Token Index: 961

Text #15

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6357. Min Act: -0.1700

Data Index: 1396104 (Python Code)

Max Activating Token Index: 576

Text #16

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6669. Min Act: -0.1700

Data Index: 1380082 (Python Code)

Max Activating Token Index: 120

Text #17

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.6044. Min Act: -0.1700

Data Index: 1370518 (Python Code)

Max Activating Token Index: 251

Text #18

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.5732. Min Act: -0.1700

Data Index: 1145321 (C4 (Web Text))

Max Activating Token Index: 620

Text #19

Max Range: 5.7919. Min Range: -5.7919

Max Act: 5.5732. Min Act: -0.1700

Data Index: 352248 (C4 (Web Text))

Max Activating Token Index: 670