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

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

Neuron 623 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: 6.4917. Min Range: -6.4917

Max Act: 6.4917. Min Act: -0.1700

Data Index: 1367458 (Python Code)

Max Activating Token Index: 205

Text #1

Max Range: 6.4917. Min Range: -6.4917

Max Act: 6.2417. Min Act: -0.1700

Data Index: 1562043 (Python Code)

Max Activating Token Index: 525

Text #2

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.8980. Min Act: -0.1700

Data Index: 1533353 (Python Code)

Max Activating Token Index: 841

Text #3

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.7730. Min Act: -0.1688

Data Index: 1175025 (C4 (Web Text))

Max Activating Token Index: 861

Text #4

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.6948. Min Act: -0.1700

Data Index: 75199 (C4 (Web Text))

Max Activating Token Index: 716

Text #5

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.7261. Min Act: -0.1700

Data Index: 312461 (C4 (Web Text))

Max Activating Token Index: 612

Text #6

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.7730. Min Act: -0.1700

Data Index: 1362814 (Python Code)

Max Activating Token Index: 289

Text #7

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.6323. Min Act: -0.1700

Data Index: 240910 (C4 (Web Text))

Max Activating Token Index: 512

Text #8

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.6480. Min Act: -0.1700

Data Index: 878835 (C4 (Web Text))

Max Activating Token Index: 136

Text #9

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.6948. Min Act: -0.1700

Data Index: 1581954 (Python Code)

Max Activating Token Index: 704

Text #10

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.5386. Min Act: -0.1700

Data Index: 1611400 (Python Code)

Max Activating Token Index: 208

Text #11

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.5386. Min Act: -0.1700

Data Index: 1593396 (Python Code)

Max Activating Token Index: 716

Text #12

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.5073. Min Act: -0.1700

Data Index: 1554966 (Python Code)

Max Activating Token Index: 488

Text #13

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.5698. Min Act: -0.1700

Data Index: 1626944 (Python Code)

Max Activating Token Index: 152

Text #14

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.5542. Min Act: -0.1700

Data Index: 1091091 (C4 (Web Text))

Max Activating Token Index: 866

Text #15

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.4448. Min Act: -0.1700

Data Index: 1523772 (Python Code)

Max Activating Token Index: 510

Text #16

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.4917. Min Act: -0.1700

Data Index: 1639136 (Python Code)

Max Activating Token Index: 925

Text #17

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.4761. Min Act: -0.1700

Data Index: 1389647 (Python Code)

Max Activating Token Index: 240

Text #18

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.4917. Min Act: -0.1700

Data Index: 1406611 (Python Code)

Max Activating Token Index: 499

Text #19

Max Range: 6.4917. Min Range: -6.4917

Max Act: 5.4448. Min Act: -0.1700

Data Index: 1597090 (Python Code)

Max Activating Token Index: 488