Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

Dataset: The Pile

Neuron 3500 in Layer 6

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l-pile')

Text #0

Max Range: 8.1972. Min Range: -8.1972

Max Act: 8.1972. Min Act: -0.0000

Data Index: 459219 (The Pile)

Max Activating Token Index: 716

Text #1

Max Range: 8.1972. Min Range: -8.1972

Max Act: 8.1742. Min Act: -0.0000

Data Index: 348913 (The Pile)

Max Activating Token Index: 201

Text #2

Max Range: 8.1972. Min Range: -8.1972

Max Act: 8.0104. Min Act: -0.0000

Data Index: 85106 (The Pile)

Max Activating Token Index: 67

Text #3

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.9544. Min Act: -0.0000

Data Index: 824250 (The Pile)

Max Activating Token Index: 955

Text #4

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.7270. Min Act: -0.0000

Data Index: 1628894 (The Pile)

Max Activating Token Index: 714

Text #5

Max Range: 8.1972. Min Range: -8.1972

Max Act: 8.0364. Min Act: -0.0000

Data Index: 1483705 (The Pile)

Max Activating Token Index: 980

Text #6

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.8021. Min Act: -0.0000

Data Index: 1400084 (The Pile)

Max Activating Token Index: 846

Text #7

Max Range: 8.1972. Min Range: -8.1972

Max Act: 8.1195. Min Act: -0.0000

Data Index: 415052 (The Pile)

Max Activating Token Index: 864

Text #8

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.8337. Min Act: -0.0000

Data Index: 261274 (The Pile)

Max Activating Token Index: 868

Text #9

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.7308. Min Act: -0.0000

Data Index: 1036206 (The Pile)

Max Activating Token Index: 762

Text #10

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.9331. Min Act: -0.0000

Data Index: 1584888 (The Pile)

Max Activating Token Index: 764

Text #11

Max Range: 8.1972. Min Range: -8.1972

Max Act: 8.0210. Min Act: -0.0000

Data Index: 565658 (The Pile)

Max Activating Token Index: 1020

Text #12

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.9173. Min Act: -0.0000

Data Index: 1649408 (The Pile)

Max Activating Token Index: 804

Text #13

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.4862. Min Act: -0.0000

Data Index: 720673 (The Pile)

Max Activating Token Index: 406

Text #14

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.2695. Min Act: -0.0000

Data Index: 1662443 (The Pile)

Max Activating Token Index: 696

Text #15

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.3343. Min Act: -0.0000

Data Index: 1730151 (The Pile)

Max Activating Token Index: 914

Text #16

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.2739. Min Act: -0.0000

Data Index: 20899 (The Pile)

Max Activating Token Index: 709

Text #17

Max Range: 8.1972. Min Range: -8.1972

Max Act: 6.7562. Min Act: -0.0000

Data Index: 1927811 (The Pile)

Max Activating Token Index: 617

Text #18

Max Range: 8.1972. Min Range: -8.1972

Max Act: 7.4428. Min Act: -0.0000

Data Index: 1638831 (The Pile)

Max Activating Token Index: 385

Text #19

Max Range: 8.1972. Min Range: -8.1972

Max Act: 6.8612. Min Act: -0.0000

Data Index: 480839 (The Pile)

Max Activating Token Index: 248