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

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

Neuron 123 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: 5.6581. Min Range: -5.6581

Max Act: 5.6581. Min Act: -0.1700

Data Index: 1039249 (C4 (Web Text))

Max Activating Token Index: 840

Text #1

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.5644. Min Act: -0.1700

Data Index: 335742 (C4 (Web Text))

Max Activating Token Index: 197

Text #2

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.4394. Min Act: -0.1700

Data Index: 263219 (C4 (Web Text))

Max Activating Token Index: 400

Text #3

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.2831. Min Act: -0.1700

Data Index: 1257427 (C4 (Web Text))

Max Activating Token Index: 331

Text #4

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.2206. Min Act: -0.1700

Data Index: 1074682 (C4 (Web Text))

Max Activating Token Index: 635

Text #5

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.2519. Min Act: -0.1700

Data Index: 173256 (C4 (Web Text))

Max Activating Token Index: 3

Text #6

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.1581. Min Act: -0.1700

Data Index: 105801 (C4 (Web Text))

Max Activating Token Index: 870

Text #7

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.1894. Min Act: -0.1700

Data Index: 1042020 (C4 (Web Text))

Max Activating Token Index: 956

Text #8

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.1581. Min Act: -0.1700

Data Index: 375767 (C4 (Web Text))

Max Activating Token Index: 191

Text #9

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.1269. Min Act: -0.1700

Data Index: 1261557 (C4 (Web Text))

Max Activating Token Index: 791

Text #10

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.0644. Min Act: -0.1700

Data Index: 214297 (C4 (Web Text))

Max Activating Token Index: 1016

Text #11

Max Range: 5.6581. Min Range: -5.6581

Max Act: 5.0019. Min Act: -0.1700

Data Index: 653069 (C4 (Web Text))

Max Activating Token Index: 891

Text #12

Max Range: 5.6581. Min Range: -5.6581

Max Act: 4.9081. Min Act: -0.1700

Data Index: 529397 (C4 (Web Text))

Max Activating Token Index: 383

Text #13

Max Range: 5.6581. Min Range: -5.6581

Max Act: 4.8456. Min Act: -0.1700

Data Index: 888928 (C4 (Web Text))

Max Activating Token Index: 882

Text #14

Max Range: 5.6581. Min Range: -5.6581

Max Act: 4.8769. Min Act: -0.1700

Data Index: 297744 (C4 (Web Text))

Max Activating Token Index: 465

Text #15

Max Range: 5.6581. Min Range: -5.6581

Max Act: 4.8456. Min Act: -0.1700

Data Index: 19053 (C4 (Web Text))

Max Activating Token Index: 763

Text #16

Max Range: 5.6581. Min Range: -5.6581

Max Act: 4.8456. Min Act: -0.1700

Data Index: 736662 (C4 (Web Text))

Max Activating Token Index: 306

Text #17

Max Range: 5.6581. Min Range: -5.6581

Max Act: 4.7519. Min Act: -0.1700

Data Index: 916864 (C4 (Web Text))

Max Activating Token Index: 234

Text #18

Max Range: 5.6581. Min Range: -5.6581

Max Act: 4.7831. Min Act: -0.1700

Data Index: 1044886 (C4 (Web Text))

Max Activating Token Index: 60

Text #19

Max Range: 5.6581. Min Range: -5.6581

Max Act: 4.7519. Min Act: -0.1699

Data Index: 389745 (C4 (Web Text))

Max Activating Token Index: 771