Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 336 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0516. Min Act: -0.0002

Data Index: 606551 (C4 (Web Text))

Max Activating Token Index: 566

Text #1

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0482. Min Act: -0.0002

Data Index: 392473 (C4 (Web Text))

Max Activating Token Index: 708

Text #2

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0470. Min Act: -0.0002

Data Index: 1499668 (Python Code)

Max Activating Token Index: 770

Text #3

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0454. Min Act: -0.0002

Data Index: 889754 (C4 (Web Text))

Max Activating Token Index: 305

Text #4

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0429. Min Act: -0.0001

Data Index: 898486 (C4 (Web Text))

Max Activating Token Index: 447

Text #5

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0457. Min Act: -0.0002

Data Index: 996618 (C4 (Web Text))

Max Activating Token Index: 226

Text #6

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0431. Min Act: -0.0002

Data Index: 1434830 (Python Code)

Max Activating Token Index: 252

Text #7

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0437. Min Act: -0.0001

Data Index: 436166 (C4 (Web Text))

Max Activating Token Index: 675

Text #8

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0423. Min Act: -0.0002

Data Index: 1328875 (C4 (Web Text))

Max Activating Token Index: 663

Text #9

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0423. Min Act: -0.0002

Data Index: 502290 (C4 (Web Text))

Max Activating Token Index: 930

Text #10

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0436. Min Act: -0.0001

Data Index: 1426122 (Python Code)

Max Activating Token Index: 402

Text #11

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0395. Min Act: -0.0001

Data Index: 243213 (C4 (Web Text))

Max Activating Token Index: 676

Text #12

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0396. Min Act: -0.0002

Data Index: 394629 (C4 (Web Text))

Max Activating Token Index: 872

Text #13

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0410. Min Act: -0.0002

Data Index: 255788 (C4 (Web Text))

Max Activating Token Index: 216

Text #14

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0417. Min Act: -0.0002

Data Index: 180136 (C4 (Web Text))

Max Activating Token Index: 796

Text #15

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0430. Min Act: -0.0001

Data Index: 480081 (C4 (Web Text))

Max Activating Token Index: 680

Text #16

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0432. Min Act: -0.0002

Data Index: 57885 (C4 (Web Text))

Max Activating Token Index: 849

Text #17

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0400. Min Act: -0.0001

Data Index: 131537 (C4 (Web Text))

Max Activating Token Index: 643

Text #18

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0381. Min Act: -0.0002

Data Index: 1060917 (C4 (Web Text))

Max Activating Token Index: 840

Text #19

Max Range: 0.0516. Min Range: -0.0516

Max Act: 0.0400. Min Act: -0.0002

Data Index: 892922 (C4 (Web Text))

Max Activating Token Index: 731