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

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

Neuron 1985 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: 3.6282. Min Range: -3.6282

Max Act: 3.6282. Min Act: -0.1700

Data Index: 1554468 (Python Code)

Max Activating Token Index: 22

Text #1

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.5029. Min Act: -0.1700

Data Index: 1446636 (Python Code)

Max Activating Token Index: 269

Text #2

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.4089. Min Act: -0.1700

Data Index: 1498832 (Python Code)

Max Activating Token Index: 525

Text #3

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.4089. Min Act: -0.1700

Data Index: 1491649 (Python Code)

Max Activating Token Index: 915

Text #4

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.3147. Min Act: -0.1700

Data Index: 769051 (C4 (Web Text))

Max Activating Token Index: 459

Text #5

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.3461. Min Act: -0.1700

Data Index: 1547104 (Python Code)

Max Activating Token Index: 702

Text #6

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2833. Min Act: -0.1700

Data Index: 1437375 (Python Code)

Max Activating Token Index: 728

Text #7

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2676. Min Act: -0.1700

Data Index: 941631 (C4 (Web Text))

Max Activating Token Index: 154

Text #8

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2676. Min Act: -0.1700

Data Index: 1541971 (Python Code)

Max Activating Token Index: 233

Text #9

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2204. Min Act: -0.1700

Data Index: 1205866 (C4 (Web Text))

Max Activating Token Index: 626

Text #10

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2519. Min Act: -0.1700

Data Index: 1493904 (Python Code)

Max Activating Token Index: 2

Text #11

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2204. Min Act: -0.1700

Data Index: 586038 (C4 (Web Text))

Max Activating Token Index: 600

Text #12

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2204. Min Act: -0.1700

Data Index: 817603 (C4 (Web Text))

Max Activating Token Index: 228

Text #13

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2047. Min Act: -0.1700

Data Index: 881824 (C4 (Web Text))

Max Activating Token Index: 48

Text #14

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.1890. Min Act: -0.1700

Data Index: 1579765 (Python Code)

Max Activating Token Index: 418

Text #15

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.1890. Min Act: -0.1700

Data Index: 442786 (C4 (Web Text))

Max Activating Token Index: 739

Text #16

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.1890. Min Act: -0.1700

Data Index: 1636007 (Python Code)

Max Activating Token Index: 78

Text #17

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2047. Min Act: -0.1700

Data Index: 1521038 (Python Code)

Max Activating Token Index: 855

Text #18

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.2362. Min Act: -0.1700

Data Index: 1651218 (Python Code)

Max Activating Token Index: 123

Text #19

Max Range: 3.6282. Min Range: -3.6282

Max Act: 3.1890. Min Act: -0.1700

Data Index: 268695 (C4 (Web Text))

Max Activating Token Index: 45