Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 5634 in Layer 11

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 1.2625. Min Range: -1.2625

Max Act: 1.2625. Min Act: -0.0000

Data Index: 83567 (C4 (Web Text))

Max Activating Token Index: 899

Text #1

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.8843. Min Act: -0.0000

Data Index: 208775 (C4 (Web Text))

Max Activating Token Index: 214

Text #2

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.7366. Min Act: -0.0000

Data Index: 514009 (C4 (Web Text))

Max Activating Token Index: 507

Text #3

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.6660. Min Act: -0.0000

Data Index: 1379681 (Python Code)

Max Activating Token Index: 352

Text #4

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.6365. Min Act: -0.0000

Data Index: 1292966 (C4 (Web Text))

Max Activating Token Index: 58

Text #5

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.5427. Min Act: 0.0000

Data Index: 809927 (C4 (Web Text))

Max Activating Token Index: 254

Text #6

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.5362. Min Act: 0.0000

Data Index: 1212208 (C4 (Web Text))

Max Activating Token Index: 790

Text #7

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4906. Min Act: 0.0000

Data Index: 1370896 (Python Code)

Max Activating Token Index: 107

Text #8

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4775. Min Act: -0.0000

Data Index: 872679 (C4 (Web Text))

Max Activating Token Index: 1002

Text #9

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4962. Min Act: -0.0000

Data Index: 1501148 (Python Code)

Max Activating Token Index: 104

Text #10

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4881. Min Act: -0.0000

Data Index: 178256 (C4 (Web Text))

Max Activating Token Index: 215

Text #11

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4200. Min Act: -0.0000

Data Index: 1514337 (Python Code)

Max Activating Token Index: 473

Text #12

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4382. Min Act: -0.0000

Data Index: 498097 (C4 (Web Text))

Max Activating Token Index: 856

Text #13

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4099. Min Act: -0.0000

Data Index: 512623 (C4 (Web Text))

Max Activating Token Index: 479

Text #14

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4106. Min Act: -0.0000

Data Index: 1602620 (Python Code)

Max Activating Token Index: 850

Text #15

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4469. Min Act: -0.0000

Data Index: 633082 (C4 (Web Text))

Max Activating Token Index: 128

Text #16

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4477. Min Act: -0.0000

Data Index: 1413661 (Python Code)

Max Activating Token Index: 982

Text #17

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4099. Min Act: -0.0000

Data Index: 1648996 (Python Code)

Max Activating Token Index: 936

Text #18

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4111. Min Act: -0.0000

Data Index: 1048043 (C4 (Web Text))

Max Activating Token Index: 895

Text #19

Max Range: 1.2625. Min Range: -1.2625

Max Act: 0.4137. Min Act: -0.0000

Data Index: 1587092 (Python Code)

Max Activating Token Index: 727