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

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

Neuron 1374 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: 5.2702. Min Range: -5.2702

Max Act: 5.2702. Min Act: -0.1700

Data Index: 1528587 (Python Code)

Max Activating Token Index: 753

Text #1

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0827. Min Act: -0.1700

Data Index: 1155051 (C4 (Web Text))

Max Activating Token Index: 395

Text #2

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.1764. Min Act: -0.1700

Data Index: 1583554 (Python Code)

Max Activating Token Index: 800

Text #3

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0827. Min Act: -0.1700

Data Index: 1014263 (C4 (Web Text))

Max Activating Token Index: 401

Text #4

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.1452. Min Act: -0.1700

Data Index: 1554370 (Python Code)

Max Activating Token Index: 302

Text #5

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.1139. Min Act: -0.1700

Data Index: 1025645 (C4 (Web Text))

Max Activating Token Index: 204

Text #6

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0827. Min Act: -0.1700

Data Index: 80174 (C4 (Web Text))

Max Activating Token Index: 39

Text #7

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0514. Min Act: -0.1700

Data Index: 730949 (C4 (Web Text))

Max Activating Token Index: 196

Text #8

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0514. Min Act: -0.1700

Data Index: 207459 (C4 (Web Text))

Max Activating Token Index: 331

Text #9

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0202. Min Act: -0.1700

Data Index: 1537625 (Python Code)

Max Activating Token Index: 428

Text #10

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0827. Min Act: -0.1700

Data Index: 487165 (C4 (Web Text))

Max Activating Token Index: 50

Text #11

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0514. Min Act: -0.1700

Data Index: 1186628 (C4 (Web Text))

Max Activating Token Index: 640

Text #12

Max Range: 5.2702. Min Range: -5.2702

Max Act: 4.9889. Min Act: -0.1700

Data Index: 1470459 (Python Code)

Max Activating Token Index: 94

Text #13

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0514. Min Act: -0.1700

Data Index: 1345967 (C4 (Web Text))

Max Activating Token Index: 563

Text #14

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0514. Min Act: -0.1700

Data Index: 515937 (C4 (Web Text))

Max Activating Token Index: 569

Text #15

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0202. Min Act: -0.1700

Data Index: 1143136 (C4 (Web Text))

Max Activating Token Index: 913

Text #16

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0514. Min Act: -0.1700

Data Index: 757652 (C4 (Web Text))

Max Activating Token Index: 292

Text #17

Max Range: 5.2702. Min Range: -5.2702

Max Act: 5.0202. Min Act: -0.1700

Data Index: 1599524 (Python Code)

Max Activating Token Index: 324

Text #18

Max Range: 5.2702. Min Range: -5.2702

Max Act: 4.9889. Min Act: -0.1699

Data Index: 1633858 (Python Code)

Max Activating Token Index: 514

Text #19

Max Range: 5.2702. Min Range: -5.2702

Max Act: 4.9264. Min Act: -0.1700

Data Index: 1613406 (Python Code)

Max Activating Token Index: 999