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

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

Neuron 1248 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: 3.7788. Min Range: -3.7788

Max Act: 3.7788. Min Act: -0.1700

Data Index: 738696 (C4 (Web Text))

Max Activating Token Index: 855

Text #1

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.7475. Min Act: -0.1700

Data Index: 898512 (C4 (Web Text))

Max Activating Token Index: 579

Text #2

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.7788. Min Act: -0.1700

Data Index: 595650 (C4 (Web Text))

Max Activating Token Index: 71

Text #3

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.7162. Min Act: -0.1700

Data Index: 1596717 (Python Code)

Max Activating Token Index: 918

Text #4

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.6849. Min Act: -0.1700

Data Index: 1020814 (C4 (Web Text))

Max Activating Token Index: 806

Text #5

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.6223. Min Act: -0.1700

Data Index: 851440 (C4 (Web Text))

Max Activating Token Index: 602

Text #6

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.5910. Min Act: -0.1700

Data Index: 1477744 (Python Code)

Max Activating Token Index: 225

Text #7

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.5284. Min Act: -0.1700

Data Index: 634731 (C4 (Web Text))

Max Activating Token Index: 951

Text #8

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.5284. Min Act: -0.1700

Data Index: 1589743 (Python Code)

Max Activating Token Index: 210

Text #9

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.5910. Min Act: -0.1700

Data Index: 1601811 (Python Code)

Max Activating Token Index: 16

Text #10

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.5597. Min Act: -0.1700

Data Index: 1309446 (C4 (Web Text))

Max Activating Token Index: 580

Text #11

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.5597. Min Act: -0.1700

Data Index: 824127 (C4 (Web Text))

Max Activating Token Index: 119

Text #12

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.5284. Min Act: -0.1700

Data Index: 433123 (C4 (Web Text))

Max Activating Token Index: 277

Text #13

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.5597. Min Act: -0.1700

Data Index: 745253 (C4 (Web Text))

Max Activating Token Index: 1016

Text #14

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.5597. Min Act: -0.1700

Data Index: 1640581 (Python Code)

Max Activating Token Index: 520

Text #15

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.4344. Min Act: -0.1700

Data Index: 753506 (C4 (Web Text))

Max Activating Token Index: 491

Text #16

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.4970. Min Act: -0.1700

Data Index: 810031 (C4 (Web Text))

Max Activating Token Index: 111

Text #17

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.4970. Min Act: -0.1700

Data Index: 1578843 (Python Code)

Max Activating Token Index: 341

Text #18

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.4970. Min Act: -0.1700

Data Index: 1228062 (C4 (Web Text))

Max Activating Token Index: 558

Text #19

Max Range: 3.7788. Min Range: -3.7788

Max Act: 3.4657. Min Act: -0.1700

Data Index: 1144297 (C4 (Web Text))

Max Activating Token Index: 510