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

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

Neuron 951 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: 5.2059. Min Range: -5.2059

Max Act: 5.2059. Min Act: -0.1700

Data Index: 376235 (C4 (Web Text))

Max Activating Token Index: 15

Text #1

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.8622. Min Act: -0.1700

Data Index: 1344663 (C4 (Web Text))

Max Activating Token Index: 341

Text #2

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.8309. Min Act: -0.1700

Data Index: 173981 (C4 (Web Text))

Max Activating Token Index: 944

Text #3

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.8622. Min Act: -0.1700

Data Index: 224646 (C4 (Web Text))

Max Activating Token Index: 519

Text #4

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.8309. Min Act: -0.1699

Data Index: 1020116 (C4 (Web Text))

Max Activating Token Index: 692

Text #5

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.7684. Min Act: -0.1700

Data Index: 1127108 (C4 (Web Text))

Max Activating Token Index: 941

Text #6

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.7216. Min Act: -0.1700

Data Index: 206012 (C4 (Web Text))

Max Activating Token Index: 829

Text #7

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.7216. Min Act: -0.1700

Data Index: 49237 (C4 (Web Text))

Max Activating Token Index: 323

Text #8

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.6903. Min Act: -0.1661

Data Index: 1313247 (C4 (Web Text))

Max Activating Token Index: 841

Text #9

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.6903. Min Act: -0.1700

Data Index: 956267 (C4 (Web Text))

Max Activating Token Index: 418

Text #10

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.7059. Min Act: -0.1700

Data Index: 665123 (C4 (Web Text))

Max Activating Token Index: 244

Text #11

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.6903. Min Act: -0.1700

Data Index: 643106 (C4 (Web Text))

Max Activating Token Index: 961

Text #12

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.6591. Min Act: -0.1700

Data Index: 726088 (C4 (Web Text))

Max Activating Token Index: 167

Text #13

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.6903. Min Act: -0.1700

Data Index: 1405881 (Python Code)

Max Activating Token Index: 883

Text #14

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.6122. Min Act: -0.1700

Data Index: 1219055 (C4 (Web Text))

Max Activating Token Index: 719

Text #15

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.6122. Min Act: -0.1700

Data Index: 519814 (C4 (Web Text))

Max Activating Token Index: 644

Text #16

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.6434. Min Act: -0.1700

Data Index: 1240583 (C4 (Web Text))

Max Activating Token Index: 439

Text #17

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.6122. Min Act: -0.1700

Data Index: 606657 (C4 (Web Text))

Max Activating Token Index: 850

Text #18

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.5653. Min Act: -0.1700

Data Index: 242404 (C4 (Web Text))

Max Activating Token Index: 655

Text #19

Max Range: 5.2059. Min Range: -5.2059

Max Act: 4.5809. Min Act: -0.1700

Data Index: 541615 (C4 (Web Text))

Max Activating Token Index: 294