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

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

Neuron 2010 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: 4.1001. Min Range: -4.1001

Max Act: 4.1001. Min Act: -0.1700

Data Index: 505713 (C4 (Web Text))

Max Activating Token Index: 381

Text #1

Max Range: 4.1001. Min Range: -4.1001

Max Act: 4.0063. Min Act: -0.1700

Data Index: 147102 (C4 (Web Text))

Max Activating Token Index: 52

Text #2

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.9594. Min Act: -0.1700

Data Index: 396427 (C4 (Web Text))

Max Activating Token Index: 266

Text #3

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.9125. Min Act: -0.1700

Data Index: 69167 (C4 (Web Text))

Max Activating Token Index: 646

Text #4

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.8968. Min Act: -0.1700

Data Index: 744863 (C4 (Web Text))

Max Activating Token Index: 931

Text #5

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.8499. Min Act: -0.1700

Data Index: 920739 (C4 (Web Text))

Max Activating Token Index: 345

Text #6

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.8343. Min Act: -0.1700

Data Index: 682813 (C4 (Web Text))

Max Activating Token Index: 370

Text #7

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.8187. Min Act: -0.1700

Data Index: 819008 (C4 (Web Text))

Max Activating Token Index: 229

Text #8

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.8499. Min Act: -0.1700

Data Index: 601081 (C4 (Web Text))

Max Activating Token Index: 631

Text #9

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.8343. Min Act: -0.1700

Data Index: 899244 (C4 (Web Text))

Max Activating Token Index: 338

Text #10

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.8030. Min Act: -0.1700

Data Index: 1354886 (C4 (Web Text))

Max Activating Token Index: 523

Text #11

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.8030. Min Act: -0.1700

Data Index: 652589 (C4 (Web Text))

Max Activating Token Index: 382

Text #12

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.8030. Min Act: -0.1700

Data Index: 243660 (C4 (Web Text))

Max Activating Token Index: 534

Text #13

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.7717. Min Act: -0.1700

Data Index: 211752 (C4 (Web Text))

Max Activating Token Index: 93

Text #14

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.7561. Min Act: -0.1700

Data Index: 593614 (C4 (Web Text))

Max Activating Token Index: 743

Text #15

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.7874. Min Act: -0.1700

Data Index: 213366 (C4 (Web Text))

Max Activating Token Index: 467

Text #16

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.7248. Min Act: -0.1700

Data Index: 526143 (C4 (Web Text))

Max Activating Token Index: 749

Text #17

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.7561. Min Act: -0.1700

Data Index: 230452 (C4 (Web Text))

Max Activating Token Index: 545

Text #18

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.7561. Min Act: -0.1700

Data Index: 221804 (C4 (Web Text))

Max Activating Token Index: 671

Text #19

Max Range: 4.1001. Min Range: -4.1001

Max Act: 3.7561. Min Act: -0.1700

Data Index: 1158279 (C4 (Web Text))

Max Activating Token Index: 776