Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 4850 in Layer 5

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1503. Min Act: -0.0000

Data Index: 1535988 (Python Code)

Max Activating Token Index: 882

Text #1

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1704. Min Act: -0.0000

Data Index: 1619333 (Python Code)

Max Activating Token Index: 282

Text #2

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1632. Min Act: -0.0000

Data Index: 1091388 (C4 (Web Text))

Max Activating Token Index: 402

Text #3

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1466. Min Act: -0.0000

Data Index: 500051 (C4 (Web Text))

Max Activating Token Index: 751

Text #4

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1484. Min Act: -0.0000

Data Index: 832943 (C4 (Web Text))

Max Activating Token Index: 804

Text #5

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1457. Min Act: -0.0000

Data Index: 1088692 (C4 (Web Text))

Max Activating Token Index: 144

Text #6

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1395. Min Act: -0.0001

Data Index: 1018080 (C4 (Web Text))

Max Activating Token Index: 504

Text #7

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1382. Min Act: -0.0000

Data Index: 1583458 (Python Code)

Max Activating Token Index: 886

Text #8

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1370. Min Act: -0.0000

Data Index: 782207 (C4 (Web Text))

Max Activating Token Index: 486

Text #9

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1355. Min Act: -0.0000

Data Index: 1100011 (C4 (Web Text))

Max Activating Token Index: 350

Text #10

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1327. Min Act: -0.0000

Data Index: 1438053 (Python Code)

Max Activating Token Index: 111

Text #11

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1244. Min Act: -0.0000

Data Index: 1204736 (C4 (Web Text))

Max Activating Token Index: 569

Text #12

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1317. Min Act: -0.0000

Data Index: 1076941 (C4 (Web Text))

Max Activating Token Index: 394

Text #13

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1127. Min Act: -0.0000

Data Index: 1401107 (Python Code)

Max Activating Token Index: 985

Text #14

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1201. Min Act: -0.0000

Data Index: 1156521 (C4 (Web Text))

Max Activating Token Index: 672

Text #15

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1236. Min Act: -0.0000

Data Index: 565718 (C4 (Web Text))

Max Activating Token Index: 347

Text #16

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1179. Min Act: -0.0000

Data Index: 878780 (C4 (Web Text))

Max Activating Token Index: 160

Text #17

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1201. Min Act: -0.0000

Data Index: 902843 (C4 (Web Text))

Max Activating Token Index: 679

Text #18

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1170. Min Act: -0.0000

Data Index: 1165351 (C4 (Web Text))

Max Activating Token Index: 213

Text #19

Max Range: 0.1704. Min Range: -0.1704

Max Act: 0.1229. Min Act: -0.0001

Data Index: 942574 (C4 (Web Text))

Max Activating Token Index: 1021