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

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

Neuron 4623 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.2002. Min Range: -3.2002

Max Act: 3.2002. Min Act: -0.0000

Data Index: 1482769 (Python Code)

Max Activating Token Index: 427

Text #1

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.7438. Min Act: -0.0001

Data Index: 1182690 (C4 (Web Text))

Max Activating Token Index: 944

Text #2

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.5078. Min Act: -0.0001

Data Index: 773887 (C4 (Web Text))

Max Activating Token Index: 292

Text #3

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.8348. Min Act: -0.0001

Data Index: 700159 (C4 (Web Text))

Max Activating Token Index: 212

Text #4

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.4183. Min Act: -0.0000

Data Index: 387166 (C4 (Web Text))

Max Activating Token Index: 18

Text #5

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.7676. Min Act: -0.0001

Data Index: 1232764 (C4 (Web Text))

Max Activating Token Index: 526

Text #6

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.8562. Min Act: -0.0001

Data Index: 334092 (C4 (Web Text))

Max Activating Token Index: 148

Text #7

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.5381. Min Act: -0.0001

Data Index: 860772 (C4 (Web Text))

Max Activating Token Index: 268

Text #8

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.5003. Min Act: -0.0000

Data Index: 1419832 (Python Code)

Max Activating Token Index: 857

Text #9

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.5353. Min Act: -0.0001

Data Index: 1124258 (C4 (Web Text))

Max Activating Token Index: 206

Text #10

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.2900. Min Act: -0.0001

Data Index: 346612 (C4 (Web Text))

Max Activating Token Index: 151

Text #11

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.1598. Min Act: -0.0000

Data Index: 788196 (C4 (Web Text))

Max Activating Token Index: 151

Text #12

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.1296. Min Act: -0.0001

Data Index: 1504869 (Python Code)

Max Activating Token Index: 442

Text #13

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.1846. Min Act: -0.0000

Data Index: 120459 (C4 (Web Text))

Max Activating Token Index: 42

Text #14

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.1495. Min Act: -0.0001

Data Index: 1448773 (Python Code)

Max Activating Token Index: 731

Text #15

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.2813. Min Act: -0.0000

Data Index: 987945 (C4 (Web Text))

Max Activating Token Index: 125

Text #16

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.2569. Min Act: -0.0000

Data Index: 1487666 (Python Code)

Max Activating Token Index: 56

Text #17

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.1925. Min Act: -0.0000

Data Index: 316106 (C4 (Web Text))

Max Activating Token Index: 111

Text #18

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.1843. Min Act: -0.0000

Data Index: 173010 (C4 (Web Text))

Max Activating Token Index: 348

Text #19

Max Range: 3.2002. Min Range: -3.2002

Max Act: 2.2840. Min Act: -0.0000

Data Index: 17610 (C4 (Web Text))

Max Activating Token Index: 172