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

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

Neuron 630 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0286. Min Act: -0.0005

Data Index: 1585179 (Python Code)

Max Activating Token Index: 852

Text #1

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0271. Min Act: -0.0005

Data Index: 698161 (C4 (Web Text))

Max Activating Token Index: 127

Text #2

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0259. Min Act: -0.0005

Data Index: 1157897 (C4 (Web Text))

Max Activating Token Index: 659

Text #3

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0263. Min Act: -0.0005

Data Index: 965914 (C4 (Web Text))

Max Activating Token Index: 535

Text #4

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0262. Min Act: -0.0005

Data Index: 1376413 (Python Code)

Max Activating Token Index: 965

Text #5

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0272. Min Act: -0.0005

Data Index: 1078443 (C4 (Web Text))

Max Activating Token Index: 673

Text #6

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0245. Min Act: -0.0005

Data Index: 442244 (C4 (Web Text))

Max Activating Token Index: 128

Text #7

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0255. Min Act: -0.0005

Data Index: 774298 (C4 (Web Text))

Max Activating Token Index: 168

Text #8

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0252. Min Act: -0.0005

Data Index: 689946 (C4 (Web Text))

Max Activating Token Index: 183

Text #9

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0255. Min Act: -0.0005

Data Index: 296553 (C4 (Web Text))

Max Activating Token Index: 371

Text #10

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0248. Min Act: -0.0005

Data Index: 1096221 (C4 (Web Text))

Max Activating Token Index: 358

Text #11

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0250. Min Act: -0.0005

Data Index: 1184169 (C4 (Web Text))

Max Activating Token Index: 388

Text #12

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0262. Min Act: -0.0005

Data Index: 858463 (C4 (Web Text))

Max Activating Token Index: 411

Text #13

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0252. Min Act: -0.0005

Data Index: 412422 (C4 (Web Text))

Max Activating Token Index: 97

Text #14

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0259. Min Act: -0.0005

Data Index: 1516731 (Python Code)

Max Activating Token Index: 305

Text #15

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0248. Min Act: -0.0005

Data Index: 390230 (C4 (Web Text))

Max Activating Token Index: 1009

Text #16

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0248. Min Act: -0.0005

Data Index: 858109 (C4 (Web Text))

Max Activating Token Index: 837

Text #17

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0233. Min Act: -0.0005

Data Index: 356871 (C4 (Web Text))

Max Activating Token Index: 798

Text #18

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0234. Min Act: -0.0005

Data Index: 1502482 (Python Code)

Max Activating Token Index: 180

Text #19

Max Range: 0.0286. Min Range: -0.0286

Max Act: 0.0247. Min Act: -0.0005

Data Index: 21148 (C4 (Web Text))

Max Activating Token Index: 978