Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 3704 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 2.1078. Min Range: -2.1078

Max Act: 2.1078. Min Act: -0.0000

Data Index: 1005709 (C4 (Web Text))

Max Activating Token Index: 986

Text #1

Max Range: 2.1078. Min Range: -2.1078

Max Act: 2.0506. Min Act: -0.0000

Data Index: 1525633 (Python Code)

Max Activating Token Index: 190

Text #2

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.9932. Min Act: -0.0000

Data Index: 1250279 (C4 (Web Text))

Max Activating Token Index: 432

Text #3

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.9217. Min Act: 0.0000

Data Index: 1208564 (C4 (Web Text))

Max Activating Token Index: 834

Text #4

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.8480. Min Act: 0.0001

Data Index: 932359 (C4 (Web Text))

Max Activating Token Index: 910

Text #5

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.8701. Min Act: -0.0000

Data Index: 1626446 (Python Code)

Max Activating Token Index: 196

Text #6

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.7065. Min Act: -0.0000

Data Index: 1642392 (Python Code)

Max Activating Token Index: 167

Text #7

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.8990. Min Act: -0.0000

Data Index: 1478024 (Python Code)

Max Activating Token Index: 543

Text #8

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.7527. Min Act: -0.0000

Data Index: 1574274 (Python Code)

Max Activating Token Index: 479

Text #9

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.9169. Min Act: -0.0000

Data Index: 522825 (C4 (Web Text))

Max Activating Token Index: 962

Text #10

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.6302. Min Act: -0.0000

Data Index: 1543913 (Python Code)

Max Activating Token Index: 152

Text #11

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.7690. Min Act: -0.0000

Data Index: 1386295 (Python Code)

Max Activating Token Index: 122

Text #12

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.6627. Min Act: -0.0000

Data Index: 1459772 (Python Code)

Max Activating Token Index: 117

Text #13

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.8504. Min Act: 0.0000

Data Index: 632207 (C4 (Web Text))

Max Activating Token Index: 980

Text #14

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.6195. Min Act: -0.0000

Data Index: 1650357 (Python Code)

Max Activating Token Index: 616

Text #15

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.6174. Min Act: -0.0000

Data Index: 924431 (C4 (Web Text))

Max Activating Token Index: 838

Text #16

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.6025. Min Act: -0.0000

Data Index: 1597217 (Python Code)

Max Activating Token Index: 451

Text #17

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.4633. Min Act: 0.0000

Data Index: 337667 (C4 (Web Text))

Max Activating Token Index: 966

Text #18

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.4412. Min Act: 0.0001

Data Index: 959991 (C4 (Web Text))

Max Activating Token Index: 579

Text #19

Max Range: 2.1078. Min Range: -2.1078

Max Act: 1.5280. Min Act: -0.0000

Data Index: 783866 (C4 (Web Text))

Max Activating Token Index: 662