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

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

Neuron 4836 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.8409. Min Act: -0.0001

Data Index: 583179 (C4 (Web Text))

Max Activating Token Index: 88

Text #1

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.7753. Min Act: -0.0001

Data Index: 303052 (C4 (Web Text))

Max Activating Token Index: 98

Text #2

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6709. Min Act: -0.0001

Data Index: 1580946 (Python Code)

Max Activating Token Index: 1009

Text #3

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.8042. Min Act: -0.0001

Data Index: 990202 (C4 (Web Text))

Max Activating Token Index: 75

Text #4

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6872. Min Act: -0.0001

Data Index: 1645674 (Python Code)

Max Activating Token Index: 32

Text #5

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6810. Min Act: -0.0001

Data Index: 978979 (C4 (Web Text))

Max Activating Token Index: 103

Text #6

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6482. Min Act: -0.0001

Data Index: 1487885 (Python Code)

Max Activating Token Index: 937

Text #7

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6609. Min Act: -0.0001

Data Index: 450801 (C4 (Web Text))

Max Activating Token Index: 96

Text #8

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6646. Min Act: -0.0001

Data Index: 1430702 (Python Code)

Max Activating Token Index: 246

Text #9

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6157. Min Act: -0.0001

Data Index: 301933 (C4 (Web Text))

Max Activating Token Index: 167

Text #10

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.5500. Min Act: -0.0001

Data Index: 458615 (C4 (Web Text))

Max Activating Token Index: 38

Text #11

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.5500. Min Act: -0.0001

Data Index: 897453 (C4 (Web Text))

Max Activating Token Index: 38

Text #12

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6061. Min Act: -0.0001

Data Index: 396250 (C4 (Web Text))

Max Activating Token Index: 53

Text #13

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6417. Min Act: -0.0001

Data Index: 963511 (C4 (Web Text))

Max Activating Token Index: 66

Text #14

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.5947. Min Act: -0.0001

Data Index: 1550534 (Python Code)

Max Activating Token Index: 8

Text #15

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.5661. Min Act: -0.0001

Data Index: 329065 (C4 (Web Text))

Max Activating Token Index: 119

Text #16

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.5130. Min Act: -0.0001

Data Index: 820538 (C4 (Web Text))

Max Activating Token Index: 257

Text #17

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.5702. Min Act: -0.0001

Data Index: 1517401 (Python Code)

Max Activating Token Index: 590

Text #18

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.5664. Min Act: -0.0001

Data Index: 1079863 (C4 (Web Text))

Max Activating Token Index: 199

Text #19

Max Range: 0.8409. Min Range: -0.8409

Max Act: 0.6293. Min Act: -0.0001

Data Index: 383743 (C4 (Web Text))

Max Activating Token Index: 343