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

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

Neuron 576 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: 5.2284. Min Range: -5.2284

Max Act: 5.2284. Min Act: -0.1700

Data Index: 538373 (C4 (Web Text))

Max Activating Token Index: 700

Text #1

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.1971. Min Act: -0.1700

Data Index: 1316381 (C4 (Web Text))

Max Activating Token Index: 622

Text #2

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.1659. Min Act: -0.1700

Data Index: 577428 (C4 (Web Text))

Max Activating Token Index: 717

Text #3

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.0721. Min Act: -0.1700

Data Index: 1165703 (C4 (Web Text))

Max Activating Token Index: 650

Text #4

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.0721. Min Act: -0.1700

Data Index: 1186744 (C4 (Web Text))

Max Activating Token Index: 739

Text #5

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.0409. Min Act: -0.1700

Data Index: 982067 (C4 (Web Text))

Max Activating Token Index: 174

Text #6

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.0721. Min Act: -0.1700

Data Index: 988040 (C4 (Web Text))

Max Activating Token Index: 906

Text #7

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.0409. Min Act: -0.1700

Data Index: 1545711 (Python Code)

Max Activating Token Index: 207

Text #8

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.0096. Min Act: -0.1700

Data Index: 935200 (C4 (Web Text))

Max Activating Token Index: 183

Text #9

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.0096. Min Act: -0.1700

Data Index: 203973 (C4 (Web Text))

Max Activating Token Index: 804

Text #10

Max Range: 5.2284. Min Range: -5.2284

Max Act: 4.9471. Min Act: -0.1700

Data Index: 968877 (C4 (Web Text))

Max Activating Token Index: 680

Text #11

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.0096. Min Act: -0.1700

Data Index: 1041753 (C4 (Web Text))

Max Activating Token Index: 879

Text #12

Max Range: 5.2284. Min Range: -5.2284

Max Act: 5.0096. Min Act: -0.1700

Data Index: 1140916 (C4 (Web Text))

Max Activating Token Index: 600

Text #13

Max Range: 5.2284. Min Range: -5.2284

Max Act: 4.9159. Min Act: -0.1700

Data Index: 1160752 (C4 (Web Text))

Max Activating Token Index: 590

Text #14

Max Range: 5.2284. Min Range: -5.2284

Max Act: 4.9471. Min Act: -0.1700

Data Index: 443528 (C4 (Web Text))

Max Activating Token Index: 616

Text #15

Max Range: 5.2284. Min Range: -5.2284

Max Act: 4.9159. Min Act: -0.1700

Data Index: 393856 (C4 (Web Text))

Max Activating Token Index: 622

Text #16

Max Range: 5.2284. Min Range: -5.2284

Max Act: 4.8221. Min Act: -0.1700

Data Index: 133273 (C4 (Web Text))

Max Activating Token Index: 839

Text #17

Max Range: 5.2284. Min Range: -5.2284

Max Act: 4.9159. Min Act: -0.1700

Data Index: 799765 (C4 (Web Text))

Max Activating Token Index: 833

Text #18

Max Range: 5.2284. Min Range: -5.2284

Max Act: 4.9159. Min Act: -0.1700

Data Index: 1205908 (C4 (Web Text))

Max Activating Token Index: 127

Text #19

Max Range: 5.2284. Min Range: -5.2284

Max Act: 4.8846. Min Act: -0.1700

Data Index: 1223684 (C4 (Web Text))

Max Activating Token Index: 637