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

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

Neuron 189 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: 5.3984. Min Range: -5.3984

Max Act: 5.3984. Min Act: -0.1700

Data Index: 380680 (C4 (Web Text))

Max Activating Token Index: 950

Text #1

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.3984. Min Act: -0.1700

Data Index: 1573542 (Python Code)

Max Activating Token Index: 15

Text #2

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.3672. Min Act: -0.1700

Data Index: 1059390 (C4 (Web Text))

Max Activating Token Index: 435

Text #3

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.3984. Min Act: -0.1700

Data Index: 944512 (C4 (Web Text))

Max Activating Token Index: 1009

Text #4

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.3359. Min Act: -0.1700

Data Index: 411594 (C4 (Web Text))

Max Activating Token Index: 632

Text #5

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.3047. Min Act: -0.1700

Data Index: 1174887 (C4 (Web Text))

Max Activating Token Index: 361

Text #6

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.3047. Min Act: -0.1700

Data Index: 184882 (C4 (Web Text))

Max Activating Token Index: 503

Text #7

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.2734. Min Act: -0.1700

Data Index: 383777 (C4 (Web Text))

Max Activating Token Index: 42

Text #8

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.2422. Min Act: -0.1700

Data Index: 385634 (C4 (Web Text))

Max Activating Token Index: 652

Text #9

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.1797. Min Act: -0.1700

Data Index: 1637956 (Python Code)

Max Activating Token Index: 513

Text #10

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.2109. Min Act: -0.1700

Data Index: 1501834 (Python Code)

Max Activating Token Index: 384

Text #11

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.2109. Min Act: -0.1700

Data Index: 662600 (C4 (Web Text))

Max Activating Token Index: 923

Text #12

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.1797. Min Act: -0.1700

Data Index: 447137 (C4 (Web Text))

Max Activating Token Index: 137

Text #13

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.1484. Min Act: -0.1700

Data Index: 423814 (C4 (Web Text))

Max Activating Token Index: 900

Text #14

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.1797. Min Act: -0.1700

Data Index: 118845 (C4 (Web Text))

Max Activating Token Index: 98

Text #15

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.1797. Min Act: -0.1700

Data Index: 706147 (C4 (Web Text))

Max Activating Token Index: 799

Text #16

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.1484. Min Act: -0.1700

Data Index: 1390460 (Python Code)

Max Activating Token Index: 595

Text #17

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.1797. Min Act: -0.1700

Data Index: 1485568 (Python Code)

Max Activating Token Index: 744

Text #18

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.1797. Min Act: -0.1700

Data Index: 1376412 (Python Code)

Max Activating Token Index: 106

Text #19

Max Range: 5.3984. Min Range: -5.3984

Max Act: 5.1797. Min Act: -0.1700

Data Index: 1362957 (Python Code)

Max Activating Token Index: 567