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

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

Neuron 382 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: 0.0298. Min Range: -0.0298

Max Act: 0.0275. Min Act: -0.0003

Data Index: 383515 (C4 (Web Text))

Max Activating Token Index: 903

Text #1

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0298. Min Act: -0.0003

Data Index: 1261164 (C4 (Web Text))

Max Activating Token Index: 738

Text #2

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0260. Min Act: -0.0002

Data Index: 1119241 (C4 (Web Text))

Max Activating Token Index: 344

Text #3

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0262. Min Act: -0.0003

Data Index: 725762 (C4 (Web Text))

Max Activating Token Index: 355

Text #4

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0253. Min Act: -0.0003

Data Index: 1495219 (Python Code)

Max Activating Token Index: 439

Text #5

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0250. Min Act: -0.0003

Data Index: 1128736 (C4 (Web Text))

Max Activating Token Index: 985

Text #6

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0251. Min Act: -0.0003

Data Index: 1322618 (C4 (Web Text))

Max Activating Token Index: 688

Text #7

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0253. Min Act: -0.0003

Data Index: 484376 (C4 (Web Text))

Max Activating Token Index: 421

Text #8

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0244. Min Act: -0.0002

Data Index: 461934 (C4 (Web Text))

Max Activating Token Index: 957

Text #9

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0243. Min Act: -0.0003

Data Index: 261827 (C4 (Web Text))

Max Activating Token Index: 24

Text #10

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0243. Min Act: -0.0003

Data Index: 19426 (C4 (Web Text))

Max Activating Token Index: 929

Text #11

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0238. Min Act: -0.0003

Data Index: 472764 (C4 (Web Text))

Max Activating Token Index: 255

Text #12

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0233. Min Act: -0.0003

Data Index: 144869 (C4 (Web Text))

Max Activating Token Index: 671

Text #13

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0231. Min Act: -0.0002

Data Index: 81194 (C4 (Web Text))

Max Activating Token Index: 84

Text #14

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0230. Min Act: -0.0002

Data Index: 765802 (C4 (Web Text))

Max Activating Token Index: 895

Text #15

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0225. Min Act: -0.0002

Data Index: 232290 (C4 (Web Text))

Max Activating Token Index: 808

Text #16

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0225. Min Act: -0.0002

Data Index: 1222750 (C4 (Web Text))

Max Activating Token Index: 882

Text #17

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0226. Min Act: -0.0003

Data Index: 806176 (C4 (Web Text))

Max Activating Token Index: 475

Text #18

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0226. Min Act: -0.0002

Data Index: 209572 (C4 (Web Text))

Max Activating Token Index: 205

Text #19

Max Range: 0.0298. Min Range: -0.0298

Max Act: 0.0224. Min Act: -0.0002

Data Index: 466397 (C4 (Web Text))

Max Activating Token Index: 273