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

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

Neuron 1951 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: 3.8137. Min Range: -3.8137

Max Act: 3.8137. Min Act: -0.1699

Data Index: 1428821 (Python Code)

Max Activating Token Index: 299

Text #1

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.7512. Min Act: -0.1700

Data Index: 1218864 (C4 (Web Text))

Max Activating Token Index: 553

Text #2

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.7199. Min Act: -0.1700

Data Index: 1486142 (Python Code)

Max Activating Token Index: 733

Text #3

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.7199. Min Act: -0.1700

Data Index: 914168 (C4 (Web Text))

Max Activating Token Index: 499

Text #4

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.6573. Min Act: -0.1700

Data Index: 1632583 (Python Code)

Max Activating Token Index: 248

Text #5

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.6416. Min Act: -0.1700

Data Index: 662945 (C4 (Web Text))

Max Activating Token Index: 893

Text #6

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.6260. Min Act: -0.1700

Data Index: 772309 (C4 (Web Text))

Max Activating Token Index: 286

Text #7

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5790. Min Act: -0.1700

Data Index: 127300 (C4 (Web Text))

Max Activating Token Index: 977

Text #8

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.6103. Min Act: -0.1700

Data Index: 329497 (C4 (Web Text))

Max Activating Token Index: 649

Text #9

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5947. Min Act: -0.1700

Data Index: 1504255 (Python Code)

Max Activating Token Index: 369

Text #10

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5947. Min Act: -0.1700

Data Index: 514118 (C4 (Web Text))

Max Activating Token Index: 940

Text #11

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5947. Min Act: -0.1700

Data Index: 451902 (C4 (Web Text))

Max Activating Token Index: 803

Text #12

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5947. Min Act: -0.1699

Data Index: 507251 (C4 (Web Text))

Max Activating Token Index: 514

Text #13

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5947. Min Act: -0.1700

Data Index: 56608 (C4 (Web Text))

Max Activating Token Index: 833

Text #14

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5790. Min Act: -0.1700

Data Index: 250764 (C4 (Web Text))

Max Activating Token Index: 331

Text #15

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5947. Min Act: -0.1700

Data Index: 169022 (C4 (Web Text))

Max Activating Token Index: 676

Text #16

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5790. Min Act: -0.1699

Data Index: 498344 (C4 (Web Text))

Max Activating Token Index: 569

Text #17

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5633. Min Act: -0.1700

Data Index: 451250 (C4 (Web Text))

Max Activating Token Index: 985

Text #18

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5947. Min Act: -0.1700

Data Index: 18785 (C4 (Web Text))

Max Activating Token Index: 864

Text #19

Max Range: 3.8137. Min Range: -3.8137

Max Act: 3.5633. Min Act: -0.1700

Data Index: 1544313 (Python Code)

Max Activating Token Index: 181