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

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

Neuron 4588 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: 2.7778. Min Range: -2.7778

Max Act: 2.7778. Min Act: -0.0000

Data Index: 1519777 (Python Code)

Max Activating Token Index: 960

Text #1

Max Range: 2.7778. Min Range: -2.7778

Max Act: 2.2957. Min Act: -0.0000

Data Index: 166762 (C4 (Web Text))

Max Activating Token Index: 109

Text #2

Max Range: 2.7778. Min Range: -2.7778

Max Act: 2.2334. Min Act: -0.0000

Data Index: 1448421 (Python Code)

Max Activating Token Index: 980

Text #3

Max Range: 2.7778. Min Range: -2.7778

Max Act: 2.2466. Min Act: -0.0000

Data Index: 1531999 (Python Code)

Max Activating Token Index: 392

Text #4

Max Range: 2.7778. Min Range: -2.7778

Max Act: 2.0351. Min Act: -0.0000

Data Index: 89902 (C4 (Web Text))

Max Activating Token Index: 175

Text #5

Max Range: 2.7778. Min Range: -2.7778

Max Act: 2.0532. Min Act: -0.0000

Data Index: 1402743 (Python Code)

Max Activating Token Index: 178

Text #6

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.9486. Min Act: -0.0000

Data Index: 330267 (C4 (Web Text))

Max Activating Token Index: 528

Text #7

Max Range: 2.7778. Min Range: -2.7778

Max Act: 2.0149. Min Act: -0.0000

Data Index: 1463217 (Python Code)

Max Activating Token Index: 339

Text #8

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.9792. Min Act: -0.0000

Data Index: 1461608 (Python Code)

Max Activating Token Index: 660

Text #9

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.9598. Min Act: -0.0000

Data Index: 1378906 (Python Code)

Max Activating Token Index: 268

Text #10

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.9377. Min Act: -0.0000

Data Index: 1195424 (C4 (Web Text))

Max Activating Token Index: 983

Text #11

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.9650. Min Act: -0.0000

Data Index: 1379382 (Python Code)

Max Activating Token Index: 651

Text #12

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.9421. Min Act: -0.0000

Data Index: 1602972 (Python Code)

Max Activating Token Index: 411

Text #13

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.9555. Min Act: -0.0000

Data Index: 1402408 (Python Code)

Max Activating Token Index: 25

Text #14

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.7566. Min Act: -0.0000

Data Index: 1644504 (Python Code)

Max Activating Token Index: 521

Text #15

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.8433. Min Act: -0.0000

Data Index: 1645144 (Python Code)

Max Activating Token Index: 944

Text #16

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.9297. Min Act: -0.0000

Data Index: 1452470 (Python Code)

Max Activating Token Index: 1013

Text #17

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.8147. Min Act: -0.0000

Data Index: 1277366 (C4 (Web Text))

Max Activating Token Index: 121

Text #18

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.8130. Min Act: -0.0000

Data Index: 1595300 (Python Code)

Max Activating Token Index: 245

Text #19

Max Range: 2.7778. Min Range: -2.7778

Max Act: 1.5971. Min Act: -0.0000

Data Index: 116426 (C4 (Web Text))

Max Activating Token Index: 220