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

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

Neuron 1865 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: 5.8798. Min Range: -5.8798

Max Act: 5.8798. Min Act: -0.1700

Data Index: 572783 (C4 (Web Text))

Max Activating Token Index: 954

Text #1

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.8798. Min Act: -0.1700

Data Index: 1219440 (C4 (Web Text))

Max Activating Token Index: 179

Text #2

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.8642. Min Act: -0.1700

Data Index: 246555 (C4 (Web Text))

Max Activating Token Index: 297

Text #3

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.7548. Min Act: -0.1700

Data Index: 1492474 (Python Code)

Max Activating Token Index: 886

Text #4

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.7235. Min Act: -0.1700

Data Index: 1204409 (C4 (Web Text))

Max Activating Token Index: 460

Text #5

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.7392. Min Act: -0.1700

Data Index: 512427 (C4 (Web Text))

Max Activating Token Index: 987

Text #6

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.7079. Min Act: -0.1700

Data Index: 367911 (C4 (Web Text))

Max Activating Token Index: 141

Text #7

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.7704. Min Act: -0.1700

Data Index: 1582275 (Python Code)

Max Activating Token Index: 753

Text #8

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.6610. Min Act: -0.1700

Data Index: 685355 (C4 (Web Text))

Max Activating Token Index: 727

Text #9

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.6767. Min Act: -0.1700

Data Index: 217869 (C4 (Web Text))

Max Activating Token Index: 883

Text #10

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.6923. Min Act: -0.1700

Data Index: 1023645 (C4 (Web Text))

Max Activating Token Index: 287

Text #11

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.6610. Min Act: -0.1700

Data Index: 495608 (C4 (Web Text))

Max Activating Token Index: 780

Text #12

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.6610. Min Act: -0.1700

Data Index: 470146 (C4 (Web Text))

Max Activating Token Index: 282

Text #13

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.6454. Min Act: -0.1700

Data Index: 16367 (C4 (Web Text))

Max Activating Token Index: 693

Text #14

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.6298. Min Act: -0.1700

Data Index: 1123602 (C4 (Web Text))

Max Activating Token Index: 157

Text #15

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.6298. Min Act: -0.1700

Data Index: 1394400 (Python Code)

Max Activating Token Index: 724

Text #16

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.6142. Min Act: -0.1700

Data Index: 1435099 (Python Code)

Max Activating Token Index: 459

Text #17

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.5829. Min Act: -0.1700

Data Index: 549906 (C4 (Web Text))

Max Activating Token Index: 125

Text #18

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.5829. Min Act: -0.1700

Data Index: 1570114 (Python Code)

Max Activating Token Index: 638

Text #19

Max Range: 5.8798. Min Range: -5.8798

Max Act: 5.5985. Min Act: -0.1700

Data Index: 316318 (C4 (Web Text))

Max Activating Token Index: 171