Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 3350 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.6936. Min Act: -0.0001

Data Index: 1451698 (Python Code)

Max Activating Token Index: 145

Text #1

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5576. Min Act: -0.0001

Data Index: 646468 (C4 (Web Text))

Max Activating Token Index: 671

Text #2

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.6220. Min Act: -0.0001

Data Index: 278247 (C4 (Web Text))

Max Activating Token Index: 131

Text #3

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.6079. Min Act: -0.0001

Data Index: 687292 (C4 (Web Text))

Max Activating Token Index: 697

Text #4

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5945. Min Act: -0.0001

Data Index: 1088857 (C4 (Web Text))

Max Activating Token Index: 332

Text #5

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5718. Min Act: -0.0001

Data Index: 101526 (C4 (Web Text))

Max Activating Token Index: 612

Text #6

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5937. Min Act: -0.0001

Data Index: 369650 (C4 (Web Text))

Max Activating Token Index: 742

Text #7

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5447. Min Act: -0.0001

Data Index: 403547 (C4 (Web Text))

Max Activating Token Index: 722

Text #8

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.6026. Min Act: -0.0001

Data Index: 201252 (C4 (Web Text))

Max Activating Token Index: 288

Text #9

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.6222. Min Act: -0.0001

Data Index: 218603 (C4 (Web Text))

Max Activating Token Index: 493

Text #10

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5758. Min Act: -0.0001

Data Index: 208073 (C4 (Web Text))

Max Activating Token Index: 101

Text #11

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5841. Min Act: -0.0001

Data Index: 140277 (C4 (Web Text))

Max Activating Token Index: 795

Text #12

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5911. Min Act: -0.0001

Data Index: 1304401 (C4 (Web Text))

Max Activating Token Index: 398

Text #13

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5711. Min Act: -0.0001

Data Index: 558391 (C4 (Web Text))

Max Activating Token Index: 367

Text #14

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5704. Min Act: -0.0001

Data Index: 1634451 (Python Code)

Max Activating Token Index: 175

Text #15

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5660. Min Act: -0.0001

Data Index: 1277171 (C4 (Web Text))

Max Activating Token Index: 16

Text #16

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5855. Min Act: -0.0001

Data Index: 1322848 (C4 (Web Text))

Max Activating Token Index: 43

Text #17

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.6254. Min Act: -0.0001

Data Index: 398698 (C4 (Web Text))

Max Activating Token Index: 802

Text #18

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5600. Min Act: -0.0001

Data Index: 1316075 (C4 (Web Text))

Max Activating Token Index: 502

Text #19

Max Range: 0.6936. Min Range: -0.6936

Max Act: 0.5569. Min Act: -0.0001

Data Index: 957319 (C4 (Web Text))

Max Activating Token Index: 498