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

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

Neuron 445 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 3.1988. Min Range: -3.1988

Max Act: 3.1988. Min Act: -0.1700

Data Index: 1103577 (C4 (Web Text))

Max Activating Token Index: 617

Text #1

Max Range: 3.1988. Min Range: -3.1988

Max Act: 3.1043. Min Act: -0.1700

Data Index: 864828 (C4 (Web Text))

Max Activating Token Index: 890

Text #2

Max Range: 3.1988. Min Range: -3.1988

Max Act: 3.1043. Min Act: -0.1700

Data Index: 534803 (C4 (Web Text))

Max Activating Token Index: 390

Text #3

Max Range: 3.1988. Min Range: -3.1988

Max Act: 3.0569. Min Act: -0.1700

Data Index: 377699 (C4 (Web Text))

Max Activating Token Index: 592

Text #4

Max Range: 3.1988. Min Range: -3.1988

Max Act: 3.0412. Min Act: -0.1700

Data Index: 105982 (C4 (Web Text))

Max Activating Token Index: 745

Text #5

Max Range: 3.1988. Min Range: -3.1988

Max Act: 3.0412. Min Act: -0.1700

Data Index: 458037 (C4 (Web Text))

Max Activating Token Index: 164

Text #6

Max Range: 3.1988. Min Range: -3.1988

Max Act: 3.0412. Min Act: -0.1700

Data Index: 24879 (C4 (Web Text))

Max Activating Token Index: 566

Text #7

Max Range: 3.1988. Min Range: -3.1988

Max Act: 3.0254. Min Act: -0.1700

Data Index: 61465 (C4 (Web Text))

Max Activating Token Index: 937

Text #8

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9938. Min Act: -0.1700

Data Index: 963264 (C4 (Web Text))

Max Activating Token Index: 54

Text #9

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9938. Min Act: -0.1700

Data Index: 125130 (C4 (Web Text))

Max Activating Token Index: 8

Text #10

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9938. Min Act: -0.1700

Data Index: 51652 (C4 (Web Text))

Max Activating Token Index: 438

Text #11

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9938. Min Act: -0.1700

Data Index: 807188 (C4 (Web Text))

Max Activating Token Index: 57

Text #12

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9938. Min Act: -0.1700

Data Index: 1131371 (C4 (Web Text))

Max Activating Token Index: 655

Text #13

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9779. Min Act: -0.1700

Data Index: 114514 (C4 (Web Text))

Max Activating Token Index: 276

Text #14

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9779. Min Act: -0.1700

Data Index: 1040967 (C4 (Web Text))

Max Activating Token Index: 354

Text #15

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9779. Min Act: -0.1700

Data Index: 1427516 (Python Code)

Max Activating Token Index: 813

Text #16

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9779. Min Act: -0.1700

Data Index: 1145077 (C4 (Web Text))

Max Activating Token Index: 724

Text #17

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9779. Min Act: -0.1700

Data Index: 838549 (C4 (Web Text))

Max Activating Token Index: 251

Text #18

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9621. Min Act: -0.1700

Data Index: 912088 (C4 (Web Text))

Max Activating Token Index: 962

Text #19

Max Range: 3.1988. Min Range: -3.1988

Max Act: 2.9779. Min Act: -0.1700

Data Index: 687397 (C4 (Web Text))

Max Activating Token Index: 272