Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 2040 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.3536. Min Act: -0.0001

Data Index: 1534316 (Python Code)

Max Activating Token Index: 268

Text #1

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.3285. Min Act: -0.0001

Data Index: 1236271 (C4 (Web Text))

Max Activating Token Index: 388

Text #2

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.3082. Min Act: -0.0001

Data Index: 1436373 (Python Code)

Max Activating Token Index: 720

Text #3

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2902. Min Act: -0.0001

Data Index: 274028 (C4 (Web Text))

Max Activating Token Index: 726

Text #4

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2968. Min Act: -0.0001

Data Index: 1288669 (C4 (Web Text))

Max Activating Token Index: 39

Text #5

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.3200. Min Act: -0.0001

Data Index: 106208 (C4 (Web Text))

Max Activating Token Index: 643

Text #6

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2917. Min Act: -0.0001

Data Index: 255393 (C4 (Web Text))

Max Activating Token Index: 732

Text #7

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2963. Min Act: -0.0001

Data Index: 348282 (C4 (Web Text))

Max Activating Token Index: 135

Text #8

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2941. Min Act: -0.0001

Data Index: 1308775 (C4 (Web Text))

Max Activating Token Index: 957

Text #9

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2814. Min Act: -0.0001

Data Index: 132251 (C4 (Web Text))

Max Activating Token Index: 478

Text #10

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2920. Min Act: -0.0001

Data Index: 1293545 (C4 (Web Text))

Max Activating Token Index: 6

Text #11

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2895. Min Act: -0.0001

Data Index: 1418842 (Python Code)

Max Activating Token Index: 325

Text #12

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2689. Min Act: -0.0001

Data Index: 1204094 (C4 (Web Text))

Max Activating Token Index: 810

Text #13

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2778. Min Act: -0.0001

Data Index: 29090 (C4 (Web Text))

Max Activating Token Index: 265

Text #14

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2763. Min Act: -0.0001

Data Index: 900719 (C4 (Web Text))

Max Activating Token Index: 1006

Text #15

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2967. Min Act: -0.0001

Data Index: 1543369 (Python Code)

Max Activating Token Index: 538

Text #16

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2755. Min Act: -0.0001

Data Index: 386942 (C4 (Web Text))

Max Activating Token Index: 13

Text #17

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2772. Min Act: -0.0001

Data Index: 1402539 (Python Code)

Max Activating Token Index: 914

Text #18

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2742. Min Act: -0.0001

Data Index: 316803 (C4 (Web Text))

Max Activating Token Index: 494

Text #19

Max Range: 0.3536. Min Range: -0.3536

Max Act: 0.2650. Min Act: -0.0001

Data Index: 771242 (C4 (Web Text))

Max Activating Token Index: 956