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

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

Neuron 1000 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.9615. Min Act: -0.0000

Data Index: 649159 (C4 (Web Text))

Max Activating Token Index: 214

Text #1

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.8532. Min Act: -0.0000

Data Index: 314434 (C4 (Web Text))

Max Activating Token Index: 977

Text #2

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.8669. Min Act: -0.0000

Data Index: 1122532 (C4 (Web Text))

Max Activating Token Index: 291

Text #3

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.8106. Min Act: -0.0000

Data Index: 463542 (C4 (Web Text))

Max Activating Token Index: 443

Text #4

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.8959. Min Act: -0.0000

Data Index: 32753 (C4 (Web Text))

Max Activating Token Index: 715

Text #5

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.8326. Min Act: -0.0000

Data Index: 989032 (C4 (Web Text))

Max Activating Token Index: 67

Text #6

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.6973. Min Act: -0.0000

Data Index: 1508695 (Python Code)

Max Activating Token Index: 145

Text #7

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.6515. Min Act: -0.0000

Data Index: 699955 (C4 (Web Text))

Max Activating Token Index: 881

Text #8

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.7345. Min Act: -0.0000

Data Index: 1435108 (Python Code)

Max Activating Token Index: 592

Text #9

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.6365. Min Act: -0.0000

Data Index: 1581792 (Python Code)

Max Activating Token Index: 350

Text #10

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.6369. Min Act: -0.0000

Data Index: 881419 (C4 (Web Text))

Max Activating Token Index: 996

Text #11

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.5987. Min Act: -0.0000

Data Index: 1390142 (Python Code)

Max Activating Token Index: 113

Text #12

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.6385. Min Act: -0.0000

Data Index: 1637319 (Python Code)

Max Activating Token Index: 1007

Text #13

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.6338. Min Act: -0.0000

Data Index: 479392 (C4 (Web Text))

Max Activating Token Index: 547

Text #14

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.6290. Min Act: -0.0000

Data Index: 278330 (C4 (Web Text))

Max Activating Token Index: 910

Text #15

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.6241. Min Act: -0.0000

Data Index: 1331555 (C4 (Web Text))

Max Activating Token Index: 49

Text #16

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.5998. Min Act: -0.0000

Data Index: 1592428 (Python Code)

Max Activating Token Index: 256

Text #17

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.5987. Min Act: -0.0000

Data Index: 449120 (C4 (Web Text))

Max Activating Token Index: 399

Text #18

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.5804. Min Act: -0.0000

Data Index: 1561159 (Python Code)

Max Activating Token Index: 60

Text #19

Max Range: 0.9615. Min Range: -0.9615

Max Act: 0.5872. Min Act: -0.0000

Data Index: 577875 (C4 (Web Text))

Max Activating Token Index: 358