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

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

Neuron 1152 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: 3.9202. Min Range: -3.9202

Max Act: 3.8577. Min Act: -0.1700

Data Index: 175980 (C4 (Web Text))

Max Activating Token Index: 938

Text #1

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.9202. Min Act: -0.1700

Data Index: 1490181 (Python Code)

Max Activating Token Index: 911

Text #2

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.8889. Min Act: -0.1700

Data Index: 149988 (C4 (Web Text))

Max Activating Token Index: 378

Text #3

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.8577. Min Act: -0.1700

Data Index: 127755 (C4 (Web Text))

Max Activating Token Index: 946

Text #4

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.9202. Min Act: -0.1700

Data Index: 650386 (C4 (Web Text))

Max Activating Token Index: 834

Text #5

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.8577. Min Act: -0.1700

Data Index: 463456 (C4 (Web Text))

Max Activating Token Index: 352

Text #6

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.8577. Min Act: -0.1700

Data Index: 519461 (C4 (Web Text))

Max Activating Token Index: 290

Text #7

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.8264. Min Act: -0.1700

Data Index: 1394820 (Python Code)

Max Activating Token Index: 845

Text #8

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.8264. Min Act: -0.1700

Data Index: 1465695 (Python Code)

Max Activating Token Index: 843

Text #9

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.7638. Min Act: -0.1700

Data Index: 1566918 (Python Code)

Max Activating Token Index: 150

Text #10

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.7951. Min Act: -0.1700

Data Index: 1492220 (Python Code)

Max Activating Token Index: 723

Text #11

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.7638. Min Act: -0.1700

Data Index: 1527942 (Python Code)

Max Activating Token Index: 408

Text #12

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.8264. Min Act: -0.1700

Data Index: 1477923 (Python Code)

Max Activating Token Index: 307

Text #13

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.7638. Min Act: -0.1700

Data Index: 1445913 (Python Code)

Max Activating Token Index: 540

Text #14

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.7638. Min Act: -0.1700

Data Index: 407660 (C4 (Web Text))

Max Activating Token Index: 763

Text #15

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.8264. Min Act: -0.1700

Data Index: 759178 (C4 (Web Text))

Max Activating Token Index: 389

Text #16

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.7638. Min Act: -0.1700

Data Index: 78894 (C4 (Web Text))

Max Activating Token Index: 54

Text #17

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.7638. Min Act: -0.1700

Data Index: 1475415 (Python Code)

Max Activating Token Index: 7

Text #18

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.7951. Min Act: -0.1700

Data Index: 1468569 (Python Code)

Max Activating Token Index: 122

Text #19

Max Range: 3.9202. Min Range: -3.9202

Max Act: 3.7325. Min Act: -0.1700

Data Index: 683872 (C4 (Web Text))

Max Activating Token Index: 121