Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 3448 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 10.1265. Min Range: -10.1265

Max Act: 10.1265. Min Act: -0.0003

Data Index: 1643779 (Python Code)

Max Activating Token Index: 706

Text #1

Max Range: 10.1265. Min Range: -10.1265

Max Act: 8.6237. Min Act: -0.0003

Data Index: 784054 (C4 (Web Text))

Max Activating Token Index: 529

Text #2

Max Range: 10.1265. Min Range: -10.1265

Max Act: 6.2306. Min Act: -0.0003

Data Index: 1173306 (C4 (Web Text))

Max Activating Token Index: 800

Text #3

Max Range: 10.1265. Min Range: -10.1265

Max Act: 5.9429. Min Act: -0.0003

Data Index: 710261 (C4 (Web Text))

Max Activating Token Index: 611

Text #4

Max Range: 10.1265. Min Range: -10.1265

Max Act: 5.5158. Min Act: -0.0003

Data Index: 34057 (C4 (Web Text))

Max Activating Token Index: 109

Text #5

Max Range: 10.1265. Min Range: -10.1265

Max Act: 5.2609. Min Act: -0.0003

Data Index: 1479881 (Python Code)

Max Activating Token Index: 172

Text #6

Max Range: 10.1265. Min Range: -10.1265

Max Act: 5.1969. Min Act: -0.0003

Data Index: 1438075 (Python Code)

Max Activating Token Index: 66

Text #7

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.9578. Min Act: -0.0003

Data Index: 1420859 (Python Code)

Max Activating Token Index: 183

Text #8

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.7790. Min Act: -0.0003

Data Index: 87020 (C4 (Web Text))

Max Activating Token Index: 403

Text #9

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.7346. Min Act: -0.0003

Data Index: 169061 (C4 (Web Text))

Max Activating Token Index: 971

Text #10

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.7046. Min Act: -0.0003

Data Index: 73637 (C4 (Web Text))

Max Activating Token Index: 722

Text #11

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.4426. Min Act: -0.0003

Data Index: 1601652 (Python Code)

Max Activating Token Index: 131

Text #12

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.4001. Min Act: -0.0004

Data Index: 483708 (C4 (Web Text))

Max Activating Token Index: 266

Text #13

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.2686. Min Act: -0.0003

Data Index: 379237 (C4 (Web Text))

Max Activating Token Index: 421

Text #14

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.3700. Min Act: -0.0003

Data Index: 1377033 (Python Code)

Max Activating Token Index: 929

Text #15

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.1355. Min Act: -0.0003

Data Index: 611447 (C4 (Web Text))

Max Activating Token Index: 635

Text #16

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.2386. Min Act: -0.0003

Data Index: 1642938 (Python Code)

Max Activating Token Index: 75

Text #17

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.2044. Min Act: -0.0003

Data Index: 637772 (C4 (Web Text))

Max Activating Token Index: 422

Text #18

Max Range: 10.1265. Min Range: -10.1265

Max Act: 4.0496. Min Act: -0.0003

Data Index: 1507671 (Python Code)

Max Activating Token Index: 125

Text #19

Max Range: 10.1265. Min Range: -10.1265

Max Act: 3.8181. Min Act: -0.0003

Data Index: 935352 (C4 (Web Text))

Max Activating Token Index: 827