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

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

Neuron 439 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.4591. Min Act: -0.1700

Data Index: 1376504 (Python Code)

Max Activating Token Index: 6

Text #1

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.4277. Min Act: -0.1700

Data Index: 1594495 (Python Code)

Max Activating Token Index: 5

Text #2

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.4591. Min Act: -0.1700

Data Index: 1624439 (Python Code)

Max Activating Token Index: 6

Text #3

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.4591. Min Act: -0.1700

Data Index: 1622512 (Python Code)

Max Activating Token Index: 6

Text #4

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.4591. Min Act: -0.1700

Data Index: 1422294 (Python Code)

Max Activating Token Index: 6

Text #5

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.4277. Min Act: -0.1700

Data Index: 1525120 (Python Code)

Max Activating Token Index: 867

Text #6

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3650. Min Act: -0.1700

Data Index: 1430296 (Python Code)

Max Activating Token Index: 404

Text #7

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3963. Min Act: -0.1700

Data Index: 1554123 (Python Code)

Max Activating Token Index: 5

Text #8

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.2707. Min Act: -0.1700

Data Index: 1484373 (Python Code)

Max Activating Token Index: 705

Text #9

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3650. Min Act: -0.1700

Data Index: 1573953 (Python Code)

Max Activating Token Index: 604

Text #10

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3336. Min Act: -0.1700

Data Index: 1491947 (Python Code)

Max Activating Token Index: 511

Text #11

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3179. Min Act: -0.1700

Data Index: 1392759 (Python Code)

Max Activating Token Index: 434

Text #12

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3493. Min Act: -0.1700

Data Index: 1459413 (Python Code)

Max Activating Token Index: 561

Text #13

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3493. Min Act: -0.1700

Data Index: 1552080 (Python Code)

Max Activating Token Index: 542

Text #14

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3022. Min Act: -0.1700

Data Index: 1601777 (Python Code)

Max Activating Token Index: 8

Text #15

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3022. Min Act: -0.1700

Data Index: 1411558 (Python Code)

Max Activating Token Index: 8

Text #16

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3179. Min Act: -0.1700

Data Index: 1514586 (Python Code)

Max Activating Token Index: 395

Text #17

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3336. Min Act: -0.1700

Data Index: 1430350 (Python Code)

Max Activating Token Index: 590

Text #18

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3179. Min Act: -0.1700

Data Index: 1439059 (Python Code)

Max Activating Token Index: 790

Text #19

Max Range: 3.4591. Min Range: -3.4591

Max Act: 3.3179. Min Act: -0.1700

Data Index: 1599000 (Python Code)

Max Activating Token Index: 616