Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 1200 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.9453. Min Act: -0.0002

Data Index: 1428876 (Python Code)

Max Activating Token Index: 770

Text #1

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.7659. Min Act: -0.0002

Data Index: 1517293 (Python Code)

Max Activating Token Index: 418

Text #2

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.7228. Min Act: -0.0002

Data Index: 1196996 (C4 (Web Text))

Max Activating Token Index: 313

Text #3

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.7001. Min Act: -0.0002

Data Index: 441134 (C4 (Web Text))

Max Activating Token Index: 338

Text #4

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6928. Min Act: -0.0002

Data Index: 1452601 (Python Code)

Max Activating Token Index: 947

Text #5

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6831. Min Act: -0.0002

Data Index: 1512087 (Python Code)

Max Activating Token Index: 638

Text #6

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.7141. Min Act: -0.0002

Data Index: 215676 (C4 (Web Text))

Max Activating Token Index: 457

Text #7

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6624. Min Act: -0.0002

Data Index: 637305 (C4 (Web Text))

Max Activating Token Index: 613

Text #8

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6633. Min Act: -0.0002

Data Index: 1214848 (C4 (Web Text))

Max Activating Token Index: 461

Text #9

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6016. Min Act: -0.0002

Data Index: 882985 (C4 (Web Text))

Max Activating Token Index: 866

Text #10

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6240. Min Act: -0.0002

Data Index: 524630 (C4 (Web Text))

Max Activating Token Index: 858

Text #11

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.5913. Min Act: -0.0002

Data Index: 287575 (C4 (Web Text))

Max Activating Token Index: 932

Text #12

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6233. Min Act: -0.0002

Data Index: 814425 (C4 (Web Text))

Max Activating Token Index: 532

Text #13

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6037. Min Act: -0.0002

Data Index: 1375457 (Python Code)

Max Activating Token Index: 829

Text #14

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6126. Min Act: -0.0002

Data Index: 1425906 (Python Code)

Max Activating Token Index: 435

Text #15

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6214. Min Act: -0.0002

Data Index: 28948 (C4 (Web Text))

Max Activating Token Index: 276

Text #16

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.5716. Min Act: -0.0002

Data Index: 503952 (C4 (Web Text))

Max Activating Token Index: 540

Text #17

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6236. Min Act: -0.0002

Data Index: 1611431 (Python Code)

Max Activating Token Index: 290

Text #18

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.6038. Min Act: -0.0002

Data Index: 1620338 (Python Code)

Max Activating Token Index: 190

Text #19

Max Range: 0.9453. Min Range: -0.9453

Max Act: 0.5876. Min Act: -0.0002

Data Index: 1149589 (C4 (Web Text))

Max Activating Token Index: 708