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

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

Neuron 636 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.6141. Min Act: -0.0000

Data Index: 306289 (C4 (Web Text))

Max Activating Token Index: 83

Text #1

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5818. Min Act: -0.0000

Data Index: 1536718 (Python Code)

Max Activating Token Index: 784

Text #2

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5141. Min Act: -0.0000

Data Index: 1653078 (Python Code)

Max Activating Token Index: 738

Text #3

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5473. Min Act: -0.0000

Data Index: 1216600 (C4 (Web Text))

Max Activating Token Index: 110

Text #4

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5196. Min Act: -0.0000

Data Index: 371511 (C4 (Web Text))

Max Activating Token Index: 97

Text #5

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5380. Min Act: -0.0000

Data Index: 1387732 (Python Code)

Max Activating Token Index: 777

Text #6

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5320. Min Act: -0.0000

Data Index: 547350 (C4 (Web Text))

Max Activating Token Index: 266

Text #7

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.4972. Min Act: -0.0000

Data Index: 1528140 (Python Code)

Max Activating Token Index: 388

Text #8

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5030. Min Act: -0.0000

Data Index: 210384 (C4 (Web Text))

Max Activating Token Index: 532

Text #9

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5061. Min Act: -0.0000

Data Index: 1222696 (C4 (Web Text))

Max Activating Token Index: 999

Text #10

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5114. Min Act: -0.0000

Data Index: 246397 (C4 (Web Text))

Max Activating Token Index: 466

Text #11

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5083. Min Act: -0.0000

Data Index: 428967 (C4 (Web Text))

Max Activating Token Index: 145

Text #12

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.5042. Min Act: -0.0000

Data Index: 737761 (C4 (Web Text))

Max Activating Token Index: 554

Text #13

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.4878. Min Act: -0.0000

Data Index: 396155 (C4 (Web Text))

Max Activating Token Index: 599

Text #14

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.4778. Min Act: -0.0000

Data Index: 1451368 (Python Code)

Max Activating Token Index: 364

Text #15

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.4893. Min Act: -0.0000

Data Index: 839405 (C4 (Web Text))

Max Activating Token Index: 43

Text #16

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.4949. Min Act: -0.0000

Data Index: 624831 (C4 (Web Text))

Max Activating Token Index: 506

Text #17

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.4132. Min Act: -0.0000

Data Index: 1371556 (Python Code)

Max Activating Token Index: 239

Text #18

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.4922. Min Act: -0.0000

Data Index: 1120665 (C4 (Web Text))

Max Activating Token Index: 94

Text #19

Max Range: 0.6141. Min Range: -0.6141

Max Act: 0.4865. Min Act: -0.0000

Data Index: 472451 (C4 (Web Text))

Max Activating Token Index: 150