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

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

Neuron 438 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 4.1624. Min Range: -4.1624

Max Act: 4.1624. Min Act: -0.1700

Data Index: 1568065 (Python Code)

Max Activating Token Index: 429

Text #1

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.8497. Min Act: -0.1700

Data Index: 1507842 (Python Code)

Max Activating Token Index: 604

Text #2

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.8184. Min Act: -0.1699

Data Index: 1151809 (C4 (Web Text))

Max Activating Token Index: 477

Text #3

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.7715. Min Act: -0.1700

Data Index: 546904 (C4 (Web Text))

Max Activating Token Index: 1023

Text #4

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.7872. Min Act: -0.1700

Data Index: 1647089 (Python Code)

Max Activating Token Index: 682

Text #5

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.7402. Min Act: -0.1700

Data Index: 512983 (C4 (Web Text))

Max Activating Token Index: 512

Text #6

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.7089. Min Act: -0.1700

Data Index: 1407389 (Python Code)

Max Activating Token Index: 897

Text #7

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.5681. Min Act: -0.1699

Data Index: 68301 (C4 (Web Text))

Max Activating Token Index: 645

Text #8

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.5837. Min Act: -0.1699

Data Index: 255360 (C4 (Web Text))

Max Activating Token Index: 714

Text #9

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.5524. Min Act: -0.1700

Data Index: 1565011 (Python Code)

Max Activating Token Index: 299

Text #10

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.5681. Min Act: -0.1700

Data Index: 371412 (C4 (Web Text))

Max Activating Token Index: 254

Text #11

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.5211. Min Act: -0.1699

Data Index: 823750 (C4 (Web Text))

Max Activating Token Index: 1023

Text #12

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.4427. Min Act: -0.1592

Data Index: 161046 (C4 (Web Text))

Max Activating Token Index: 572

Text #13

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.3643. Min Act: -0.1699

Data Index: 1016910 (C4 (Web Text))

Max Activating Token Index: 419

Text #14

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.3957. Min Act: -0.1699

Data Index: 844851 (C4 (Web Text))

Max Activating Token Index: 435

Text #15

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.3800. Min Act: -0.1700

Data Index: 1474466 (Python Code)

Max Activating Token Index: 132

Text #16

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.3800. Min Act: -0.1700

Data Index: 1592816 (Python Code)

Max Activating Token Index: 150

Text #17

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.3486. Min Act: -0.1700

Data Index: 964163 (C4 (Web Text))

Max Activating Token Index: 400

Text #18

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.3329. Min Act: -0.1700

Data Index: 576330 (C4 (Web Text))

Max Activating Token Index: 495

Text #19

Max Range: 4.1624. Min Range: -4.1624

Max Act: 3.3172. Min Act: -0.1700

Data Index: 1281877 (C4 (Web Text))

Max Activating Token Index: 512