Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 512 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0423. Min Act: -0.0006

Data Index: 1638207 (Python Code)

Max Activating Token Index: 662

Text #1

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0402. Min Act: -0.0006

Data Index: 1421547 (Python Code)

Max Activating Token Index: 710

Text #2

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0384. Min Act: -0.0006

Data Index: 390274 (C4 (Web Text))

Max Activating Token Index: 24

Text #3

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0363. Min Act: -0.0006

Data Index: 1038961 (C4 (Web Text))

Max Activating Token Index: 264

Text #4

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0359. Min Act: -0.0006

Data Index: 1117551 (C4 (Web Text))

Max Activating Token Index: 100

Text #5

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0350. Min Act: -0.0006

Data Index: 510404 (C4 (Web Text))

Max Activating Token Index: 907

Text #6

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0356. Min Act: -0.0006

Data Index: 450392 (C4 (Web Text))

Max Activating Token Index: 32

Text #7

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0359. Min Act: -0.0006

Data Index: 947054 (C4 (Web Text))

Max Activating Token Index: 104

Text #8

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0351. Min Act: -0.0006

Data Index: 235581 (C4 (Web Text))

Max Activating Token Index: 44

Text #9

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0365. Min Act: -0.0006

Data Index: 1491428 (Python Code)

Max Activating Token Index: 637

Text #10

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0371. Min Act: -0.0006

Data Index: 1512893 (Python Code)

Max Activating Token Index: 572

Text #11

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0340. Min Act: -0.0006

Data Index: 95165 (C4 (Web Text))

Max Activating Token Index: 529

Text #12

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0320. Min Act: -0.0006

Data Index: 1393033 (Python Code)

Max Activating Token Index: 500

Text #13

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0350. Min Act: -0.0006

Data Index: 1181865 (C4 (Web Text))

Max Activating Token Index: 229

Text #14

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0346. Min Act: -0.0006

Data Index: 1364401 (Python Code)

Max Activating Token Index: 997

Text #15

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0353. Min Act: -0.0006

Data Index: 634230 (C4 (Web Text))

Max Activating Token Index: 338

Text #16

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0345. Min Act: -0.0006

Data Index: 647978 (C4 (Web Text))

Max Activating Token Index: 27

Text #17

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0332. Min Act: -0.0006

Data Index: 1525680 (Python Code)

Max Activating Token Index: 441

Text #18

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0342. Min Act: -0.0006

Data Index: 244080 (C4 (Web Text))

Max Activating Token Index: 134

Text #19

Max Range: 0.0423. Min Range: -0.0423

Max Act: 0.0343. Min Act: -0.0006

Data Index: 499771 (C4 (Web Text))

Max Activating Token Index: 238