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

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

Neuron 2018 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')

Text #0

Max Range: 5.2231. Min Range: -5.2231

Max Act: 5.2231. Min Act: -0.1700

Data Index: 426659 (C4 (Web Text))

Max Activating Token Index: 1023

Text #1

Max Range: 5.2231. Min Range: -5.2231

Max Act: 5.1919. Min Act: -0.1700

Data Index: 594834 (C4 (Web Text))

Max Activating Token Index: 1023

Text #2

Max Range: 5.2231. Min Range: -5.2231

Max Act: 5.0356. Min Act: -0.1700

Data Index: 745213 (C4 (Web Text))

Max Activating Token Index: 1023

Text #3

Max Range: 5.2231. Min Range: -5.2231

Max Act: 5.0356. Min Act: -0.1700

Data Index: 1126386 (C4 (Web Text))

Max Activating Token Index: 1023

Text #4

Max Range: 5.2231. Min Range: -5.2231

Max Act: 5.0044. Min Act: -0.1700

Data Index: 944136 (C4 (Web Text))

Max Activating Token Index: 1023

Text #5

Max Range: 5.2231. Min Range: -5.2231

Max Act: 5.0044. Min Act: -0.1700

Data Index: 180438 (C4 (Web Text))

Max Activating Token Index: 1023

Text #6

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.9419. Min Act: -0.1700

Data Index: 1338593 (C4 (Web Text))

Max Activating Token Index: 1023

Text #7

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.9731. Min Act: -0.1700

Data Index: 328882 (C4 (Web Text))

Max Activating Token Index: 1023

Text #8

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.9106. Min Act: -0.1700

Data Index: 718995 (C4 (Web Text))

Max Activating Token Index: 1023

Text #9

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.9106. Min Act: -0.1700

Data Index: 90294 (C4 (Web Text))

Max Activating Token Index: 1023

Text #10

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.8481. Min Act: -0.1700

Data Index: 1018396 (C4 (Web Text))

Max Activating Token Index: 1023

Text #11

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.8169. Min Act: -0.1700

Data Index: 771694 (C4 (Web Text))

Max Activating Token Index: 1023

Text #12

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.7543. Min Act: -0.1700

Data Index: 1358352 (C4 (Web Text))

Max Activating Token Index: 1023

Text #13

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.7231. Min Act: -0.1700

Data Index: 158782 (C4 (Web Text))

Max Activating Token Index: 1023

Text #14

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.7856. Min Act: -0.1700

Data Index: 768383 (C4 (Web Text))

Max Activating Token Index: 1023

Text #15

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.6606. Min Act: -0.1700

Data Index: 1020888 (C4 (Web Text))

Max Activating Token Index: 1023

Text #16

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.5356. Min Act: -0.1700

Data Index: 1617230 (Python Code)

Max Activating Token Index: 1023

Text #17

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.4731. Min Act: -0.1700

Data Index: 184389 (C4 (Web Text))

Max Activating Token Index: 1023

Text #18

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.5043. Min Act: -0.1700

Data Index: 905804 (C4 (Web Text))

Max Activating Token Index: 1023

Text #19

Max Range: 5.2231. Min Range: -5.2231

Max Act: 4.4418. Min Act: -0.1700

Data Index: 809729 (C4 (Web Text))

Max Activating Token Index: 1023