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

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

Neuron 4071 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.6930. Min Act: -0.0000

Data Index: 25938 (C4 (Web Text))

Max Activating Token Index: 405

Text #1

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.7326. Min Act: -0.0000

Data Index: 1502180 (Python Code)

Max Activating Token Index: 911

Text #2

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.6350. Min Act: -0.0000

Data Index: 1209534 (C4 (Web Text))

Max Activating Token Index: 460

Text #3

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.6765. Min Act: -0.0000

Data Index: 76348 (C4 (Web Text))

Max Activating Token Index: 372

Text #4

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.6297. Min Act: -0.0000

Data Index: 479476 (C4 (Web Text))

Max Activating Token Index: 10

Text #5

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.6439. Min Act: -0.0000

Data Index: 1528354 (Python Code)

Max Activating Token Index: 314

Text #6

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5991. Min Act: -0.0000

Data Index: 443492 (C4 (Web Text))

Max Activating Token Index: 249

Text #7

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.6289. Min Act: -0.0000

Data Index: 1046696 (C4 (Web Text))

Max Activating Token Index: 278

Text #8

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5731. Min Act: -0.0000

Data Index: 60559 (C4 (Web Text))

Max Activating Token Index: 571

Text #9

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5737. Min Act: -0.0000

Data Index: 287535 (C4 (Web Text))

Max Activating Token Index: 147

Text #10

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5973. Min Act: -0.0000

Data Index: 1001391 (C4 (Web Text))

Max Activating Token Index: 695

Text #11

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5733. Min Act: -0.0000

Data Index: 1528574 (Python Code)

Max Activating Token Index: 630

Text #12

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5938. Min Act: -0.0000

Data Index: 587693 (C4 (Web Text))

Max Activating Token Index: 1018

Text #13

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5533. Min Act: -0.0000

Data Index: 1527587 (Python Code)

Max Activating Token Index: 19

Text #14

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.6059. Min Act: -0.0000

Data Index: 130528 (C4 (Web Text))

Max Activating Token Index: 586

Text #15

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5376. Min Act: -0.0000

Data Index: 1275770 (C4 (Web Text))

Max Activating Token Index: 920

Text #16

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5243. Min Act: -0.0001

Data Index: 1286625 (C4 (Web Text))

Max Activating Token Index: 608

Text #17

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5343. Min Act: -0.0000

Data Index: 1236428 (C4 (Web Text))

Max Activating Token Index: 13

Text #18

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5455. Min Act: -0.0000

Data Index: 510022 (C4 (Web Text))

Max Activating Token Index: 94

Text #19

Max Range: 0.7326. Min Range: -0.7326

Max Act: 0.5714. Min Act: -0.0000

Data Index: 1599172 (Python Code)

Max Activating Token Index: 502