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

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

Neuron 1630 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: 7.4874. Min Range: -7.4874

Max Act: 7.4874. Min Act: -0.1699

Data Index: 1516751 (Python Code)

Max Activating Token Index: 741

Text #1

Max Range: 7.4874. Min Range: -7.4874

Max Act: 7.2999. Min Act: -0.1699

Data Index: 606829 (C4 (Web Text))

Max Activating Token Index: 607

Text #2

Max Range: 7.4874. Min Range: -7.4874

Max Act: 7.1436. Min Act: -0.1699

Data Index: 425316 (C4 (Web Text))

Max Activating Token Index: 885

Text #3

Max Range: 7.4874. Min Range: -7.4874

Max Act: 7.1280. Min Act: -0.1699

Data Index: 1032926 (C4 (Web Text))

Max Activating Token Index: 910

Text #4

Max Range: 7.4874. Min Range: -7.4874

Max Act: 7.0968. Min Act: -0.1699

Data Index: 699037 (C4 (Web Text))

Max Activating Token Index: 1022

Text #5

Max Range: 7.4874. Min Range: -7.4874

Max Act: 7.0811. Min Act: -0.1699

Data Index: 1202074 (C4 (Web Text))

Max Activating Token Index: 405

Text #6

Max Range: 7.4874. Min Range: -7.4874

Max Act: 7.0343. Min Act: -0.1699

Data Index: 1059480 (C4 (Web Text))

Max Activating Token Index: 1001

Text #7

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.9093. Min Act: -0.1699

Data Index: 1572407 (Python Code)

Max Activating Token Index: 192

Text #8

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7843. Min Act: -0.1699

Data Index: 1234066 (C4 (Web Text))

Max Activating Token Index: 620

Text #9

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7843. Min Act: -0.1699

Data Index: 1461947 (Python Code)

Max Activating Token Index: 877

Text #10

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7530. Min Act: -0.1699

Data Index: 1409703 (Python Code)

Max Activating Token Index: 448

Text #11

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7530. Min Act: -0.1699

Data Index: 1618277 (Python Code)

Max Activating Token Index: 715

Text #12

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7218. Min Act: -0.1699

Data Index: 294595 (C4 (Web Text))

Max Activating Token Index: 598

Text #13

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7218. Min Act: -0.1699

Data Index: 300490 (C4 (Web Text))

Max Activating Token Index: 294

Text #14

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.8311. Min Act: -0.1699

Data Index: 1442176 (Python Code)

Max Activating Token Index: 925

Text #15

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7843. Min Act: -0.1699

Data Index: 1194518 (C4 (Web Text))

Max Activating Token Index: 717

Text #16

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7061. Min Act: -0.1699

Data Index: 1532670 (Python Code)

Max Activating Token Index: 684

Text #17

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7061. Min Act: -0.1699

Data Index: 528730 (C4 (Web Text))

Max Activating Token Index: 775

Text #18

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.7061. Min Act: -0.1699

Data Index: 229897 (C4 (Web Text))

Max Activating Token Index: 802

Text #19

Max Range: 7.4874. Min Range: -7.4874

Max Act: 6.6436. Min Act: -0.1699

Data Index: 1575034 (Python Code)

Max Activating Token Index: 651