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

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

Neuron 341 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.8426. Min Act: -0.1700

Data Index: 704374 (C4 (Web Text))

Max Activating Token Index: 131

Text #1

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.6239. Min Act: -0.1700

Data Index: 420355 (C4 (Web Text))

Max Activating Token Index: 141

Text #2

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.4676. Min Act: -0.1700

Data Index: 1644262 (Python Code)

Max Activating Token Index: 252

Text #3

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.4364. Min Act: -0.1700

Data Index: 1028244 (C4 (Web Text))

Max Activating Token Index: 215

Text #4

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.2489. Min Act: -0.1700

Data Index: 790413 (C4 (Web Text))

Max Activating Token Index: 927

Text #5

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.2176. Min Act: -0.1700

Data Index: 1574348 (Python Code)

Max Activating Token Index: 429

Text #6

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.2489. Min Act: -0.1700

Data Index: 312757 (C4 (Web Text))

Max Activating Token Index: 731

Text #7

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.1551. Min Act: -0.1700

Data Index: 1549980 (Python Code)

Max Activating Token Index: 260

Text #8

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.1864. Min Act: -0.1700

Data Index: 1444010 (Python Code)

Max Activating Token Index: 817

Text #9

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.1551. Min Act: -0.1700

Data Index: 1606971 (Python Code)

Max Activating Token Index: 707

Text #10

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.1239. Min Act: -0.1700

Data Index: 251858 (C4 (Web Text))

Max Activating Token Index: 333

Text #11

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.0614. Min Act: -0.1700

Data Index: 1609574 (Python Code)

Max Activating Token Index: 481

Text #12

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.0926. Min Act: -0.1700

Data Index: 111590 (C4 (Web Text))

Max Activating Token Index: 331

Text #13

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.0301. Min Act: -0.1700

Data Index: 1553893 (Python Code)

Max Activating Token Index: 114

Text #14

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.0301. Min Act: -0.1700

Data Index: 315018 (C4 (Web Text))

Max Activating Token Index: 151

Text #15

Max Range: 5.8426. Min Range: -5.8426

Max Act: 5.0614. Min Act: -0.1700

Data Index: 334370 (C4 (Web Text))

Max Activating Token Index: 60

Text #16

Max Range: 5.8426. Min Range: -5.8426

Max Act: 4.9676. Min Act: -0.1700

Data Index: 445090 (C4 (Web Text))

Max Activating Token Index: 154

Text #17

Max Range: 5.8426. Min Range: -5.8426

Max Act: 4.9989. Min Act: -0.1700

Data Index: 1490494 (Python Code)

Max Activating Token Index: 647

Text #18

Max Range: 5.8426. Min Range: -5.8426

Max Act: 4.9989. Min Act: -0.1700

Data Index: 423756 (C4 (Web Text))

Max Activating Token Index: 197

Text #19

Max Range: 5.8426. Min Range: -5.8426

Max Act: 4.9364. Min Act: -0.1700

Data Index: 1474089 (Python Code)

Max Activating Token Index: 865