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

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

Neuron 1312 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: 3.8850. Min Range: -3.8850

Max Act: 3.8850. Min Act: -0.1700

Data Index: 1108900 (C4 (Web Text))

Max Activating Token Index: 112

Text #1

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.8850. Min Act: -0.1700

Data Index: 663412 (C4 (Web Text))

Max Activating Token Index: 695

Text #2

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.8537. Min Act: -0.1700

Data Index: 1555999 (Python Code)

Max Activating Token Index: 707

Text #3

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.7912. Min Act: -0.1700

Data Index: 1562134 (Python Code)

Max Activating Token Index: 515

Text #4

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.7599. Min Act: -0.1700

Data Index: 1583181 (Python Code)

Max Activating Token Index: 87

Text #5

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.7286. Min Act: -0.1700

Data Index: 865196 (C4 (Web Text))

Max Activating Token Index: 212

Text #6

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.7286. Min Act: -0.1700

Data Index: 1376402 (Python Code)

Max Activating Token Index: 781

Text #7

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.7286. Min Act: -0.1700

Data Index: 129088 (C4 (Web Text))

Max Activating Token Index: 484

Text #8

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6973. Min Act: -0.1700

Data Index: 817340 (C4 (Web Text))

Max Activating Token Index: 465

Text #9

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6973. Min Act: -0.1700

Data Index: 1469634 (Python Code)

Max Activating Token Index: 176

Text #10

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6660. Min Act: -0.1700

Data Index: 1609630 (Python Code)

Max Activating Token Index: 273

Text #11

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6660. Min Act: -0.1700

Data Index: 1520272 (Python Code)

Max Activating Token Index: 48

Text #12

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.7286. Min Act: -0.1700

Data Index: 1453948 (Python Code)

Max Activating Token Index: 727

Text #13

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6660. Min Act: -0.1700

Data Index: 149067 (C4 (Web Text))

Max Activating Token Index: 630

Text #14

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6347. Min Act: -0.1700

Data Index: 260270 (C4 (Web Text))

Max Activating Token Index: 367

Text #15

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6347. Min Act: -0.1700

Data Index: 1653793 (Python Code)

Max Activating Token Index: 48

Text #16

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6973. Min Act: -0.1700

Data Index: 1642980 (Python Code)

Max Activating Token Index: 619

Text #17

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6347. Min Act: -0.1700

Data Index: 1631198 (Python Code)

Max Activating Token Index: 92

Text #18

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6347. Min Act: -0.1700

Data Index: 1544963 (Python Code)

Max Activating Token Index: 391

Text #19

Max Range: 3.8850. Min Range: -3.8850

Max Act: 3.6660. Min Act: -0.1700

Data Index: 1466895 (Python Code)

Max Activating Token Index: 643