Model: SoLU Model: 2 Layers, 2048 Neurons per Layer

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

Neuron 27 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-2l')

Text #0

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0453. Min Act: -0.0003

Data Index: 308759 (C4 (Web Text))

Max Activating Token Index: 957

Text #1

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0361. Min Act: -0.0003

Data Index: 1035307 (C4 (Web Text))

Max Activating Token Index: 943

Text #2

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0312. Min Act: -0.0004

Data Index: 601173 (C4 (Web Text))

Max Activating Token Index: 137

Text #3

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0297. Min Act: -0.0003

Data Index: 1442739 (Python Code)

Max Activating Token Index: 1008

Text #4

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0284. Min Act: -0.0003

Data Index: 1514781 (Python Code)

Max Activating Token Index: 15

Text #5

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0281. Min Act: -0.0003

Data Index: 1261746 (C4 (Web Text))

Max Activating Token Index: 736

Text #6

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0288. Min Act: -0.0003

Data Index: 1580182 (Python Code)

Max Activating Token Index: 286

Text #7

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0292. Min Act: -0.0003

Data Index: 1074237 (C4 (Web Text))

Max Activating Token Index: 1009

Text #8

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0271. Min Act: -0.0002

Data Index: 1586198 (Python Code)

Max Activating Token Index: 98

Text #9

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0275. Min Act: -0.0003

Data Index: 19660 (C4 (Web Text))

Max Activating Token Index: 327

Text #10

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0272. Min Act: -0.0003

Data Index: 1492370 (Python Code)

Max Activating Token Index: 35

Text #11

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0272. Min Act: -0.0003

Data Index: 1574981 (Python Code)

Max Activating Token Index: 106

Text #12

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0268. Min Act: -0.0002

Data Index: 1410965 (Python Code)

Max Activating Token Index: 116

Text #13

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0257. Min Act: -0.0003

Data Index: 772897 (C4 (Web Text))

Max Activating Token Index: 803

Text #14

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0267. Min Act: -0.0003

Data Index: 1596039 (Python Code)

Max Activating Token Index: 137

Text #15

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0248. Min Act: -0.0003

Data Index: 1578259 (Python Code)

Max Activating Token Index: 12

Text #16

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0249. Min Act: -0.0003

Data Index: 1323510 (C4 (Web Text))

Max Activating Token Index: 853

Text #17

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0248. Min Act: -0.0002

Data Index: 695379 (C4 (Web Text))

Max Activating Token Index: 42

Text #18

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0239. Min Act: -0.0003

Data Index: 1172134 (C4 (Web Text))

Max Activating Token Index: 663

Text #19

Max Range: 0.0453. Min Range: -0.0453

Max Act: 0.0234. Min Act: -0.0003

Data Index: 277034 (C4 (Web Text))

Max Activating Token Index: 62