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

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

Neuron 1177 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 7.0402. Min Range: -7.0402

Max Act: 7.0402. Min Act: -0.1700

Data Index: 1412164 (Python Code)

Max Activating Token Index: 454

Text #1

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.6027. Min Act: -0.1700

Data Index: 1006761 (C4 (Web Text))

Max Activating Token Index: 409

Text #2

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.4777. Min Act: -0.1700

Data Index: 883220 (C4 (Web Text))

Max Activating Token Index: 523

Text #3

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.4777. Min Act: -0.1700

Data Index: 1177719 (C4 (Web Text))

Max Activating Token Index: 234

Text #4

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.2277. Min Act: -0.1700

Data Index: 436681 (C4 (Web Text))

Max Activating Token Index: 482

Text #5

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.4152. Min Act: -0.1700

Data Index: 989967 (C4 (Web Text))

Max Activating Token Index: 854

Text #6

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.2277. Min Act: -0.1700

Data Index: 26666 (C4 (Web Text))

Max Activating Token Index: 270

Text #7

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.2902. Min Act: -0.1700

Data Index: 1041031 (C4 (Web Text))

Max Activating Token Index: 686

Text #8

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.4777. Min Act: -0.1700

Data Index: 1051704 (C4 (Web Text))

Max Activating Token Index: 105

Text #9

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.2902. Min Act: -0.1700

Data Index: 1615986 (Python Code)

Max Activating Token Index: 310

Text #10

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.1027. Min Act: -0.1700

Data Index: 261295 (C4 (Web Text))

Max Activating Token Index: 752

Text #11

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.2902. Min Act: -0.1700

Data Index: 1163507 (C4 (Web Text))

Max Activating Token Index: 398

Text #12

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.1652. Min Act: -0.1700

Data Index: 309684 (C4 (Web Text))

Max Activating Token Index: 350

Text #13

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.1652. Min Act: -0.1699

Data Index: 1335994 (C4 (Web Text))

Max Activating Token Index: 917

Text #14

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.2277. Min Act: -0.1700

Data Index: 120559 (C4 (Web Text))

Max Activating Token Index: 158

Text #15

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.4152. Min Act: -0.1700

Data Index: 1430082 (Python Code)

Max Activating Token Index: 947

Text #16

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.2277. Min Act: -0.1700

Data Index: 1026663 (C4 (Web Text))

Max Activating Token Index: 781

Text #17

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.1027. Min Act: -0.1700

Data Index: 329953 (C4 (Web Text))

Max Activating Token Index: 291

Text #18

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.2277. Min Act: -0.1700

Data Index: 160277 (C4 (Web Text))

Max Activating Token Index: 671

Text #19

Max Range: 7.0402. Min Range: -7.0402

Max Act: 6.1652. Min Act: -0.1699

Data Index: 1003615 (C4 (Web Text))

Max Activating Token Index: 518