Model: GELU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 81 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.7680. Min Act: -0.1700

Data Index: 1461397 (Python Code)

Max Activating Token Index: 277

Text #1

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.7055. Min Act: -0.1700

Data Index: 1319079 (C4 (Web Text))

Max Activating Token Index: 886

Text #2

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.7211. Min Act: -0.1700

Data Index: 1512137 (Python Code)

Max Activating Token Index: 919

Text #3

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.7211. Min Act: -0.1700

Data Index: 1399387 (Python Code)

Max Activating Token Index: 150

Text #4

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.6742. Min Act: -0.1700

Data Index: 1495930 (Python Code)

Max Activating Token Index: 896

Text #5

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.6742. Min Act: -0.1700

Data Index: 1011466 (C4 (Web Text))

Max Activating Token Index: 157

Text #6

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.6429. Min Act: -0.1700

Data Index: 1332991 (C4 (Web Text))

Max Activating Token Index: 166

Text #7

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.5959. Min Act: -0.1700

Data Index: 1570249 (Python Code)

Max Activating Token Index: 301

Text #8

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.5802. Min Act: -0.1700

Data Index: 1508723 (Python Code)

Max Activating Token Index: 642

Text #9

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.6272. Min Act: -0.1700

Data Index: 30870 (C4 (Web Text))

Max Activating Token Index: 152

Text #10

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.6116. Min Act: -0.1700

Data Index: 1070859 (C4 (Web Text))

Max Activating Token Index: 180

Text #11

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.5646. Min Act: -0.1700

Data Index: 1394859 (Python Code)

Max Activating Token Index: 242

Text #12

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.6116. Min Act: -0.1700

Data Index: 866259 (C4 (Web Text))

Max Activating Token Index: 835

Text #13

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.5802. Min Act: -0.1700

Data Index: 220673 (C4 (Web Text))

Max Activating Token Index: 297

Text #14

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.6585. Min Act: -0.1700

Data Index: 1639407 (Python Code)

Max Activating Token Index: 472

Text #15

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.6116. Min Act: -0.1700

Data Index: 1581530 (Python Code)

Max Activating Token Index: 678

Text #16

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.5802. Min Act: -0.1700

Data Index: 1312306 (C4 (Web Text))

Max Activating Token Index: 141

Text #17

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.5646. Min Act: -0.1700

Data Index: 482650 (C4 (Web Text))

Max Activating Token Index: 616

Text #18

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.6116. Min Act: -0.1700

Data Index: 1616104 (Python Code)

Max Activating Token Index: 624

Text #19

Max Range: 3.7680. Min Range: -3.7680

Max Act: 3.5489. Min Act: -0.1700

Data Index: 1611619 (Python Code)

Max Activating Token Index: 292