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

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

Neuron 92 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.3658. Min Act: -0.1700

Data Index: 618721 (C4 (Web Text))

Max Activating Token Index: 245

Text #1

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.3344. Min Act: -0.1700

Data Index: 884114 (C4 (Web Text))

Max Activating Token Index: 475

Text #2

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.2559. Min Act: -0.1700

Data Index: 1541512 (Python Code)

Max Activating Token Index: 113

Text #3

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.2559. Min Act: -0.1700

Data Index: 1060292 (C4 (Web Text))

Max Activating Token Index: 368

Text #4

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.2402. Min Act: -0.1700

Data Index: 616282 (C4 (Web Text))

Max Activating Token Index: 523

Text #5

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.2245. Min Act: -0.1700

Data Index: 580726 (C4 (Web Text))

Max Activating Token Index: 302

Text #6

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.2402. Min Act: -0.1700

Data Index: 307627 (C4 (Web Text))

Max Activating Token Index: 323

Text #7

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.2402. Min Act: -0.1700

Data Index: 929769 (C4 (Web Text))

Max Activating Token Index: 219

Text #8

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.1773. Min Act: -0.1700

Data Index: 866720 (C4 (Web Text))

Max Activating Token Index: 466

Text #9

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.2245. Min Act: -0.1700

Data Index: 896971 (C4 (Web Text))

Max Activating Token Index: 208

Text #10

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.1615. Min Act: -0.1700

Data Index: 1561389 (Python Code)

Max Activating Token Index: 330

Text #11

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.2402. Min Act: -0.1700

Data Index: 273060 (C4 (Web Text))

Max Activating Token Index: 415

Text #12

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.1458. Min Act: -0.1700

Data Index: 1458056 (Python Code)

Max Activating Token Index: 173

Text #13

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.2087. Min Act: -0.1700

Data Index: 1519754 (Python Code)

Max Activating Token Index: 655

Text #14

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.1773. Min Act: -0.1700

Data Index: 1578928 (Python Code)

Max Activating Token Index: 832

Text #15

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.1300. Min Act: -0.1700

Data Index: 80880 (C4 (Web Text))

Max Activating Token Index: 305

Text #16

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.1458. Min Act: -0.1700

Data Index: 1104469 (C4 (Web Text))

Max Activating Token Index: 122

Text #17

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.1773. Min Act: -0.1700

Data Index: 993017 (C4 (Web Text))

Max Activating Token Index: 224

Text #18

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.1143. Min Act: -0.1700

Data Index: 1441760 (Python Code)

Max Activating Token Index: 700

Text #19

Max Range: 3.3658. Min Range: -3.3658

Max Act: 3.1615. Min Act: -0.1700

Data Index: 391943 (C4 (Web Text))

Max Activating Token Index: 127