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

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

Neuron 1228 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: 5.6534. Min Range: -5.6534

Max Act: 5.6534. Min Act: -0.1700

Data Index: 1017234 (C4 (Web Text))

Max Activating Token Index: 507

Text #1

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.5909. Min Act: -0.1700

Data Index: 1448101 (Python Code)

Max Activating Token Index: 1021

Text #2

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.4659. Min Act: -0.1700

Data Index: 1643790 (Python Code)

Max Activating Token Index: 959

Text #3

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.4346. Min Act: -0.1700

Data Index: 1599227 (Python Code)

Max Activating Token Index: 455

Text #4

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.4659. Min Act: -0.1700

Data Index: 1411069 (Python Code)

Max Activating Token Index: 424

Text #5

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.3721. Min Act: -0.1700

Data Index: 1153292 (C4 (Web Text))

Max Activating Token Index: 176

Text #6

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.3721. Min Act: -0.1700

Data Index: 849526 (C4 (Web Text))

Max Activating Token Index: 442

Text #7

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.3409. Min Act: -0.1700

Data Index: 1600863 (Python Code)

Max Activating Token Index: 711

Text #8

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.3409. Min Act: -0.1700

Data Index: 473656 (C4 (Web Text))

Max Activating Token Index: 333

Text #9

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.2784. Min Act: -0.1700

Data Index: 165173 (C4 (Web Text))

Max Activating Token Index: 174

Text #10

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.4034. Min Act: -0.1700

Data Index: 1639129 (Python Code)

Max Activating Token Index: 517

Text #11

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.3409. Min Act: -0.1700

Data Index: 1545644 (Python Code)

Max Activating Token Index: 534

Text #12

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.1846. Min Act: -0.1700

Data Index: 996363 (C4 (Web Text))

Max Activating Token Index: 197

Text #13

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.2159. Min Act: -0.1700

Data Index: 1081252 (C4 (Web Text))

Max Activating Token Index: 776

Text #14

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.2471. Min Act: -0.1700

Data Index: 423543 (C4 (Web Text))

Max Activating Token Index: 161

Text #15

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.2159. Min Act: -0.1700

Data Index: 1592168 (Python Code)

Max Activating Token Index: 140

Text #16

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.2471. Min Act: -0.1700

Data Index: 1086487 (C4 (Web Text))

Max Activating Token Index: 254

Text #17

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.1846. Min Act: -0.1700

Data Index: 590604 (C4 (Web Text))

Max Activating Token Index: 616

Text #18

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.2159. Min Act: -0.1700

Data Index: 90739 (C4 (Web Text))

Max Activating Token Index: 257

Text #19

Max Range: 5.6534. Min Range: -5.6534

Max Act: 5.2471. Min Act: -0.1700

Data Index: 1119541 (C4 (Web Text))

Max Activating Token Index: 269