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

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

Neuron 1898 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 4.1994. Min Range: -4.1994

Max Act: 4.1994. Min Act: -0.1700

Data Index: 846219 (C4 (Web Text))

Max Activating Token Index: 463

Text #1

Max Range: 4.1994. Min Range: -4.1994

Max Act: 4.1369. Min Act: -0.1700

Data Index: 1303652 (C4 (Web Text))

Max Activating Token Index: 550

Text #2

Max Range: 4.1994. Min Range: -4.1994

Max Act: 4.0431. Min Act: -0.1700

Data Index: 269280 (C4 (Web Text))

Max Activating Token Index: 820

Text #3

Max Range: 4.1994. Min Range: -4.1994

Max Act: 4.0118. Min Act: -0.1700

Data Index: 768360 (C4 (Web Text))

Max Activating Token Index: 310

Text #4

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.8868. Min Act: -0.1700

Data Index: 152966 (C4 (Web Text))

Max Activating Token Index: 673

Text #5

Max Range: 4.1994. Min Range: -4.1994

Max Act: 4.0118. Min Act: -0.1700

Data Index: 20950 (C4 (Web Text))

Max Activating Token Index: 607

Text #6

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.9180. Min Act: -0.1700

Data Index: 64847 (C4 (Web Text))

Max Activating Token Index: 515

Text #7

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.8242. Min Act: -0.1700

Data Index: 459980 (C4 (Web Text))

Max Activating Token Index: 660

Text #8

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.9180. Min Act: -0.1700

Data Index: 998316 (C4 (Web Text))

Max Activating Token Index: 172

Text #9

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.8555. Min Act: -0.1700

Data Index: 1286120 (C4 (Web Text))

Max Activating Token Index: 714

Text #10

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.8555. Min Act: -0.1700

Data Index: 320866 (C4 (Web Text))

Max Activating Token Index: 615

Text #11

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.7929. Min Act: -0.1700

Data Index: 1483940 (Python Code)

Max Activating Token Index: 861

Text #12

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.8555. Min Act: -0.1700

Data Index: 636307 (C4 (Web Text))

Max Activating Token Index: 665

Text #13

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.8868. Min Act: -0.1700

Data Index: 922164 (C4 (Web Text))

Max Activating Token Index: 708

Text #14

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.8555. Min Act: -0.1700

Data Index: 107219 (C4 (Web Text))

Max Activating Token Index: 668

Text #15

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.7929. Min Act: -0.1700

Data Index: 60519 (C4 (Web Text))

Max Activating Token Index: 598

Text #16

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.7929. Min Act: -0.1700

Data Index: 1354497 (C4 (Web Text))

Max Activating Token Index: 686

Text #17

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.7929. Min Act: -0.1700

Data Index: 859522 (C4 (Web Text))

Max Activating Token Index: 882

Text #18

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.8242. Min Act: -0.1700

Data Index: 93069 (C4 (Web Text))

Max Activating Token Index: 487

Text #19

Max Range: 4.1994. Min Range: -4.1994

Max Act: 3.7929. Min Act: -0.1700

Data Index: 271290 (C4 (Web Text))

Max Activating Token Index: 976