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

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

Neuron 1458 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: 5.9141. Min Range: -5.9141

Max Act: 5.9141. Min Act: -0.1700

Data Index: 1469700 (Python Code)

Max Activating Token Index: 602

Text #1

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.7266. Min Act: -0.1700

Data Index: 1393090 (Python Code)

Max Activating Token Index: 1007

Text #2

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.5703. Min Act: -0.1700

Data Index: 113116 (C4 (Web Text))

Max Activating Token Index: 513

Text #3

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.6016. Min Act: -0.1700

Data Index: 1436834 (Python Code)

Max Activating Token Index: 838

Text #4

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.5391. Min Act: -0.1700

Data Index: 1100612 (C4 (Web Text))

Max Activating Token Index: 414

Text #5

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.5703. Min Act: -0.1697

Data Index: 1595211 (Python Code)

Max Activating Token Index: 108

Text #6

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.5391. Min Act: -0.1700

Data Index: 1593082 (Python Code)

Max Activating Token Index: 141

Text #7

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.5078. Min Act: -0.1700

Data Index: 617917 (C4 (Web Text))

Max Activating Token Index: 500

Text #8

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.4766. Min Act: -0.1700

Data Index: 683003 (C4 (Web Text))

Max Activating Token Index: 947

Text #9

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.4453. Min Act: -0.1700

Data Index: 1471515 (Python Code)

Max Activating Token Index: 857

Text #10

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.5078. Min Act: -0.1700

Data Index: 1452644 (Python Code)

Max Activating Token Index: 996

Text #11

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.4453. Min Act: -0.1700

Data Index: 762986 (C4 (Web Text))

Max Activating Token Index: 513

Text #12

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.4453. Min Act: -0.1700

Data Index: 1500185 (Python Code)

Max Activating Token Index: 161

Text #13

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.4141. Min Act: -0.1700

Data Index: 1367356 (Python Code)

Max Activating Token Index: 241

Text #14

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.3828. Min Act: -0.1700

Data Index: 1645442 (Python Code)

Max Activating Token Index: 819

Text #15

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.3828. Min Act: -0.1693

Data Index: 1474937 (Python Code)

Max Activating Token Index: 136

Text #16

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.3828. Min Act: -0.1700

Data Index: 1430683 (Python Code)

Max Activating Token Index: 353

Text #17

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.4141. Min Act: -0.1700

Data Index: 1502556 (Python Code)

Max Activating Token Index: 724

Text #18

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.3828. Min Act: -0.1699

Data Index: 1380896 (Python Code)

Max Activating Token Index: 910

Text #19

Max Range: 5.9141. Min Range: -5.9141

Max Act: 5.4141. Min Act: -0.1700

Data Index: 1472473 (Python Code)

Max Activating Token Index: 88