Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 1600 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0358. Min Act: -0.0003

Data Index: 1121691 (C4 (Web Text))

Max Activating Token Index: 55

Text #1

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0343. Min Act: -0.0003

Data Index: 795874 (C4 (Web Text))

Max Activating Token Index: 142

Text #2

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0323. Min Act: -0.0003

Data Index: 1001245 (C4 (Web Text))

Max Activating Token Index: 382

Text #3

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0316. Min Act: -0.0003

Data Index: 1213864 (C4 (Web Text))

Max Activating Token Index: 737

Text #4

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0303. Min Act: -0.0004

Data Index: 151313 (C4 (Web Text))

Max Activating Token Index: 760

Text #5

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0307. Min Act: -0.0003

Data Index: 673335 (C4 (Web Text))

Max Activating Token Index: 837

Text #6

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0294. Min Act: -0.0003

Data Index: 154004 (C4 (Web Text))

Max Activating Token Index: 458

Text #7

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0275. Min Act: -0.0003

Data Index: 602692 (C4 (Web Text))

Max Activating Token Index: 270

Text #8

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0273. Min Act: -0.0003

Data Index: 1553116 (Python Code)

Max Activating Token Index: 418

Text #9

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0273. Min Act: -0.0003

Data Index: 811886 (C4 (Web Text))

Max Activating Token Index: 150

Text #10

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0281. Min Act: -0.0003

Data Index: 836100 (C4 (Web Text))

Max Activating Token Index: 204

Text #11

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0275. Min Act: -0.0003

Data Index: 1309080 (C4 (Web Text))

Max Activating Token Index: 55

Text #12

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0286. Min Act: -0.0003

Data Index: 723538 (C4 (Web Text))

Max Activating Token Index: 200

Text #13

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0261. Min Act: -0.0003

Data Index: 1112865 (C4 (Web Text))

Max Activating Token Index: 795

Text #14

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0269. Min Act: -0.0003

Data Index: 980268 (C4 (Web Text))

Max Activating Token Index: 845

Text #15

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0269. Min Act: -0.0003

Data Index: 585694 (C4 (Web Text))

Max Activating Token Index: 461

Text #16

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0259. Min Act: -0.0003

Data Index: 300539 (C4 (Web Text))

Max Activating Token Index: 32

Text #17

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0268. Min Act: -0.0003

Data Index: 1338386 (C4 (Web Text))

Max Activating Token Index: 261

Text #18

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0268. Min Act: -0.0003

Data Index: 1375714 (Python Code)

Max Activating Token Index: 282

Text #19

Max Range: 0.0358. Min Range: -0.0358

Max Act: 0.0252. Min Act: -0.0003

Data Index: 252057 (C4 (Web Text))

Max Activating Token Index: 70