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

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

Neuron 28 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: 4.1105. Min Range: -4.1105

Max Act: 4.1105. Min Act: -0.1700

Data Index: 817155 (C4 (Web Text))

Max Activating Token Index: 171

Text #1

Max Range: 4.1105. Min Range: -4.1105

Max Act: 4.0948. Min Act: -0.1700

Data Index: 1089196 (C4 (Web Text))

Max Activating Token Index: 197

Text #2

Max Range: 4.1105. Min Range: -4.1105

Max Act: 4.0792. Min Act: -0.1700

Data Index: 937910 (C4 (Web Text))

Max Activating Token Index: 433

Text #3

Max Range: 4.1105. Min Range: -4.1105

Max Act: 4.0323. Min Act: -0.1700

Data Index: 866787 (C4 (Web Text))

Max Activating Token Index: 696

Text #4

Max Range: 4.1105. Min Range: -4.1105

Max Act: 4.0167. Min Act: -0.1700

Data Index: 632548 (C4 (Web Text))

Max Activating Token Index: 746

Text #5

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.9541. Min Act: -0.1700

Data Index: 290855 (C4 (Web Text))

Max Activating Token Index: 556

Text #6

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.9229. Min Act: -0.1700

Data Index: 752152 (C4 (Web Text))

Max Activating Token Index: 714

Text #7

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8760. Min Act: -0.1700

Data Index: 949492 (C4 (Web Text))

Max Activating Token Index: 133

Text #8

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8760. Min Act: -0.1700

Data Index: 975215 (C4 (Web Text))

Max Activating Token Index: 341

Text #9

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8447. Min Act: -0.1700

Data Index: 996024 (C4 (Web Text))

Max Activating Token Index: 80

Text #10

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8603. Min Act: -0.1700

Data Index: 600461 (C4 (Web Text))

Max Activating Token Index: 770

Text #11

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8291. Min Act: -0.1700

Data Index: 653229 (C4 (Web Text))

Max Activating Token Index: 557

Text #12

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8291. Min Act: -0.1700

Data Index: 520633 (C4 (Web Text))

Max Activating Token Index: 206

Text #13

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8291. Min Act: -0.1700

Data Index: 617345 (C4 (Web Text))

Max Activating Token Index: 417

Text #14

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8447. Min Act: -0.1700

Data Index: 900034 (C4 (Web Text))

Max Activating Token Index: 998

Text #15

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8291. Min Act: -0.1700

Data Index: 872932 (C4 (Web Text))

Max Activating Token Index: 690

Text #16

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8134. Min Act: -0.1700

Data Index: 907791 (C4 (Web Text))

Max Activating Token Index: 707

Text #17

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8291. Min Act: -0.1700

Data Index: 486833 (C4 (Web Text))

Max Activating Token Index: 794

Text #18

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.8134. Min Act: -0.1700

Data Index: 402493 (C4 (Web Text))

Max Activating Token Index: 814

Text #19

Max Range: 4.1105. Min Range: -4.1105

Max Act: 3.7665. Min Act: -0.1700

Data Index: 5653 (C4 (Web Text))

Max Activating Token Index: 762