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

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

Neuron 995 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.2580. Min Act: -0.1700

Data Index: 798278 (C4 (Web Text))

Max Activating Token Index: 286

Text #1

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.2893. Min Act: -0.1700

Data Index: 717027 (C4 (Web Text))

Max Activating Token Index: 337

Text #2

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.3830. Min Act: -0.1700

Data Index: 668687 (C4 (Web Text))

Max Activating Token Index: 849

Text #3

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.2580. Min Act: -0.1700

Data Index: 77831 (C4 (Web Text))

Max Activating Token Index: 58

Text #4

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.2580. Min Act: -0.1700

Data Index: 96022 (C4 (Web Text))

Max Activating Token Index: 443

Text #5

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.2893. Min Act: -0.1700

Data Index: 1350917 (C4 (Web Text))

Max Activating Token Index: 559

Text #6

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.1330. Min Act: -0.1700

Data Index: 1437723 (Python Code)

Max Activating Token Index: 499

Text #7

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.1018. Min Act: -0.1700

Data Index: 1431627 (Python Code)

Max Activating Token Index: 144

Text #8

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.1018. Min Act: -0.1700

Data Index: 441168 (C4 (Web Text))

Max Activating Token Index: 322

Text #9

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.1330. Min Act: -0.1700

Data Index: 1240259 (C4 (Web Text))

Max Activating Token Index: 865

Text #10

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.1018. Min Act: -0.1700

Data Index: 981944 (C4 (Web Text))

Max Activating Token Index: 911

Text #11

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.1018. Min Act: -0.1700

Data Index: 1471772 (Python Code)

Max Activating Token Index: 352

Text #12

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.0080. Min Act: -0.1700

Data Index: 1237255 (C4 (Web Text))

Max Activating Token Index: 750

Text #13

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.0705. Min Act: -0.1700

Data Index: 38141 (C4 (Web Text))

Max Activating Token Index: 366

Text #14

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.0705. Min Act: -0.1700

Data Index: 559636 (C4 (Web Text))

Max Activating Token Index: 842

Text #15

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.0080. Min Act: -0.1700

Data Index: 719208 (C4 (Web Text))

Max Activating Token Index: 254

Text #16

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.0080. Min Act: -0.1700

Data Index: 1238251 (C4 (Web Text))

Max Activating Token Index: 175

Text #17

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.0080. Min Act: -0.1700

Data Index: 1462782 (Python Code)

Max Activating Token Index: 1019

Text #18

Max Range: 5.3830. Min Range: -5.3830

Max Act: 4.9455. Min Act: -0.1700

Data Index: 1636128 (Python Code)

Max Activating Token Index: 642

Text #19

Max Range: 5.3830. Min Range: -5.3830

Max Act: 5.0705. Min Act: -0.1700

Data Index: 497753 (C4 (Web Text))

Max Activating Token Index: 622