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

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

Neuron 3871 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.9198. Min Act: -0.0001

Data Index: 1245531 (C4 (Web Text))

Max Activating Token Index: 304

Text #1

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.8073. Min Act: -0.0001

Data Index: 1472316 (Python Code)

Max Activating Token Index: 918

Text #2

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.9142. Min Act: -0.0001

Data Index: 1565813 (Python Code)

Max Activating Token Index: 122

Text #3

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.7935. Min Act: -0.0001

Data Index: 673850 (C4 (Web Text))

Max Activating Token Index: 145

Text #4

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.8206. Min Act: -0.0001

Data Index: 1353077 (C4 (Web Text))

Max Activating Token Index: 478

Text #5

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.8212. Min Act: -0.0001

Data Index: 1254576 (C4 (Web Text))

Max Activating Token Index: 214

Text #6

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.8092. Min Act: -0.0001

Data Index: 1539374 (Python Code)

Max Activating Token Index: 14

Text #7

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.8403. Min Act: -0.0001

Data Index: 238580 (C4 (Web Text))

Max Activating Token Index: 681

Text #8

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.8038. Min Act: -0.0001

Data Index: 119533 (C4 (Web Text))

Max Activating Token Index: 534

Text #9

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.8081. Min Act: -0.0001

Data Index: 941761 (C4 (Web Text))

Max Activating Token Index: 92

Text #10

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.7882. Min Act: -0.0001

Data Index: 1498576 (Python Code)

Max Activating Token Index: 42

Text #11

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.7722. Min Act: -0.0001

Data Index: 1470247 (Python Code)

Max Activating Token Index: 778

Text #12

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.7267. Min Act: -0.0001

Data Index: 1503288 (Python Code)

Max Activating Token Index: 685

Text #13

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.7388. Min Act: -0.0001

Data Index: 113326 (C4 (Web Text))

Max Activating Token Index: 305

Text #14

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.7441. Min Act: -0.0001

Data Index: 494200 (C4 (Web Text))

Max Activating Token Index: 540

Text #15

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.7407. Min Act: -0.0001

Data Index: 1247224 (C4 (Web Text))

Max Activating Token Index: 253

Text #16

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.6515. Min Act: -0.0001

Data Index: 1639620 (Python Code)

Max Activating Token Index: 101

Text #17

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.7386. Min Act: -0.0001

Data Index: 272603 (C4 (Web Text))

Max Activating Token Index: 407

Text #18

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.7914. Min Act: -0.0001

Data Index: 1465325 (Python Code)

Max Activating Token Index: 417

Text #19

Max Range: 0.9198. Min Range: -0.9198

Max Act: 0.6740. Min Act: -0.0001

Data Index: 1209660 (C4 (Web Text))

Max Activating Token Index: 205