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

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

Neuron 1511 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: 5.9158. Min Range: -5.9158

Max Act: 5.9158. Min Act: -0.1700

Data Index: 1457351 (Python Code)

Max Activating Token Index: 175

Text #1

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.8845. Min Act: -0.1700

Data Index: 571233 (C4 (Web Text))

Max Activating Token Index: 331

Text #2

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.8220. Min Act: -0.1700

Data Index: 791978 (C4 (Web Text))

Max Activating Token Index: 756

Text #3

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.7908. Min Act: -0.1700

Data Index: 1263116 (C4 (Web Text))

Max Activating Token Index: 471

Text #4

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.7283. Min Act: -0.1700

Data Index: 126732 (C4 (Web Text))

Max Activating Token Index: 704

Text #5

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.7283. Min Act: -0.1700

Data Index: 532329 (C4 (Web Text))

Max Activating Token Index: 793

Text #6

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.7283. Min Act: -0.1700

Data Index: 209937 (C4 (Web Text))

Max Activating Token Index: 473

Text #7

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.6970. Min Act: -0.1700

Data Index: 1060300 (C4 (Web Text))

Max Activating Token Index: 671

Text #8

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.6658. Min Act: -0.1700

Data Index: 603975 (C4 (Web Text))

Max Activating Token Index: 955

Text #9

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.7283. Min Act: -0.1700

Data Index: 292108 (C4 (Web Text))

Max Activating Token Index: 946

Text #10

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.6970. Min Act: -0.1700

Data Index: 1182014 (C4 (Web Text))

Max Activating Token Index: 742

Text #11

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.6345. Min Act: -0.1700

Data Index: 261258 (C4 (Web Text))

Max Activating Token Index: 454

Text #12

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.6658. Min Act: -0.1700

Data Index: 1045721 (C4 (Web Text))

Max Activating Token Index: 466

Text #13

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.6033. Min Act: -0.1700

Data Index: 1540334 (Python Code)

Max Activating Token Index: 378

Text #14

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.6345. Min Act: -0.1700

Data Index: 1003629 (C4 (Web Text))

Max Activating Token Index: 163

Text #15

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.6658. Min Act: -0.1700

Data Index: 74492 (C4 (Web Text))

Max Activating Token Index: 426

Text #16

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.5720. Min Act: -0.1700

Data Index: 312549 (C4 (Web Text))

Max Activating Token Index: 534

Text #17

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.6345. Min Act: -0.1700

Data Index: 318893 (C4 (Web Text))

Max Activating Token Index: 33

Text #18

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.5720. Min Act: -0.1700

Data Index: 392940 (C4 (Web Text))

Max Activating Token Index: 629

Text #19

Max Range: 5.9158. Min Range: -5.9158

Max Act: 5.5408. Min Act: -0.1700

Data Index: 1149934 (C4 (Web Text))

Max Activating Token Index: 591