Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 4347 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.3078. Min Act: -0.0000

Data Index: 477042 (C4 (Web Text))

Max Activating Token Index: 658

Text #1

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.3071. Min Act: -0.0000

Data Index: 840926 (C4 (Web Text))

Max Activating Token Index: 140

Text #2

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2745. Min Act: -0.0000

Data Index: 1206541 (C4 (Web Text))

Max Activating Token Index: 367

Text #3

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2951. Min Act: -0.0000

Data Index: 908655 (C4 (Web Text))

Max Activating Token Index: 455

Text #4

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2755. Min Act: -0.0001

Data Index: 586815 (C4 (Web Text))

Max Activating Token Index: 191

Text #5

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2638. Min Act: -0.0000

Data Index: 730515 (C4 (Web Text))

Max Activating Token Index: 513

Text #6

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2518. Min Act: -0.0000

Data Index: 560788 (C4 (Web Text))

Max Activating Token Index: 938

Text #7

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2513. Min Act: -0.0000

Data Index: 270288 (C4 (Web Text))

Max Activating Token Index: 1005

Text #8

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2609. Min Act: -0.0000

Data Index: 583305 (C4 (Web Text))

Max Activating Token Index: 138

Text #9

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2449. Min Act: -0.0000

Data Index: 1031724 (C4 (Web Text))

Max Activating Token Index: 1010

Text #10

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2344. Min Act: -0.0000

Data Index: 121312 (C4 (Web Text))

Max Activating Token Index: 280

Text #11

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2334. Min Act: -0.0000

Data Index: 1617932 (Python Code)

Max Activating Token Index: 203

Text #12

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2343. Min Act: -0.0001

Data Index: 827615 (C4 (Web Text))

Max Activating Token Index: 1011

Text #13

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2305. Min Act: -0.0000

Data Index: 1269298 (C4 (Web Text))

Max Activating Token Index: 257

Text #14

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2375. Min Act: -0.0000

Data Index: 97316 (C4 (Web Text))

Max Activating Token Index: 240

Text #15

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2390. Min Act: -0.0000

Data Index: 448883 (C4 (Web Text))

Max Activating Token Index: 897

Text #16

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2307. Min Act: -0.0000

Data Index: 439927 (C4 (Web Text))

Max Activating Token Index: 252

Text #17

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2265. Min Act: -0.0001

Data Index: 280151 (C4 (Web Text))

Max Activating Token Index: 592

Text #18

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2434. Min Act: -0.0000

Data Index: 1186990 (C4 (Web Text))

Max Activating Token Index: 653

Text #19

Max Range: 0.3078. Min Range: -0.3078

Max Act: 0.2345. Min Act: -0.0000

Data Index: 2692 (C4 (Web Text))

Max Activating Token Index: 200