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

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

Neuron 608 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: 7.7353. Min Range: -7.7353

Max Act: 7.7353. Min Act: -0.1700

Data Index: 87202 (C4 (Web Text))

Max Activating Token Index: 444

Text #1

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.6728. Min Act: -0.1700

Data Index: 1411795 (Python Code)

Max Activating Token Index: 713

Text #2

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.4853. Min Act: -0.1700

Data Index: 1565377 (Python Code)

Max Activating Token Index: 845

Text #3

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.5478. Min Act: -0.1700

Data Index: 1131489 (C4 (Web Text))

Max Activating Token Index: 956

Text #4

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.4853. Min Act: -0.1700

Data Index: 1537537 (Python Code)

Max Activating Token Index: 332

Text #5

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.4853. Min Act: -0.1700

Data Index: 1196996 (C4 (Web Text))

Max Activating Token Index: 196

Text #6

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.4228. Min Act: -0.1700

Data Index: 441134 (C4 (Web Text))

Max Activating Token Index: 196

Text #7

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.3603. Min Act: -0.1700

Data Index: 1498571 (Python Code)

Max Activating Token Index: 375

Text #8

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.4228. Min Act: -0.1700

Data Index: 1607127 (Python Code)

Max Activating Token Index: 309

Text #9

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.2978. Min Act: -0.1700

Data Index: 1385217 (Python Code)

Max Activating Token Index: 818

Text #10

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.2978. Min Act: -0.1700

Data Index: 677386 (C4 (Web Text))

Max Activating Token Index: 667

Text #11

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.2353. Min Act: -0.1700

Data Index: 1467791 (Python Code)

Max Activating Token Index: 326

Text #12

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.2978. Min Act: -0.1700

Data Index: 205348 (C4 (Web Text))

Max Activating Token Index: 513

Text #13

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.2353. Min Act: -0.1700

Data Index: 565297 (C4 (Web Text))

Max Activating Token Index: 710

Text #14

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.2353. Min Act: -0.1700

Data Index: 1508391 (Python Code)

Max Activating Token Index: 995

Text #15

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.2353. Min Act: -0.1700

Data Index: 1095828 (C4 (Web Text))

Max Activating Token Index: 1016

Text #16

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.1728. Min Act: -0.1700

Data Index: 1370504 (Python Code)

Max Activating Token Index: 547

Text #17

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.2353. Min Act: -0.1700

Data Index: 1454470 (Python Code)

Max Activating Token Index: 507

Text #18

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.1728. Min Act: -0.1700

Data Index: 1585845 (Python Code)

Max Activating Token Index: 925

Text #19

Max Range: 7.7353. Min Range: -7.7353

Max Act: 7.1728. Min Act: -0.1700

Data Index: 1214848 (C4 (Web Text))

Max Activating Token Index: 320