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

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

Neuron 1537 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.7425. Min Act: -0.1700

Data Index: 488276 (C4 (Web Text))

Max Activating Token Index: 951

Text #1

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.5550. Min Act: -0.1700

Data Index: 334408 (C4 (Web Text))

Max Activating Token Index: 823

Text #2

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.6175. Min Act: -0.1700

Data Index: 1367969 (Python Code)

Max Activating Token Index: 147

Text #3

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.6175. Min Act: -0.1700

Data Index: 31964 (C4 (Web Text))

Max Activating Token Index: 942

Text #4

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.5550. Min Act: -0.1700

Data Index: 91306 (C4 (Web Text))

Max Activating Token Index: 42

Text #5

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.5550. Min Act: -0.1700

Data Index: 1408580 (Python Code)

Max Activating Token Index: 192

Text #6

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.5550. Min Act: -0.1700

Data Index: 668347 (C4 (Web Text))

Max Activating Token Index: 638

Text #7

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.4925. Min Act: -0.1700

Data Index: 264256 (C4 (Web Text))

Max Activating Token Index: 368

Text #8

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.4300. Min Act: -0.1700

Data Index: 1234920 (C4 (Web Text))

Max Activating Token Index: 964

Text #9

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.3675. Min Act: -0.1700

Data Index: 612627 (C4 (Web Text))

Max Activating Token Index: 724

Text #10

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.4613. Min Act: -0.1700

Data Index: 59323 (C4 (Web Text))

Max Activating Token Index: 972

Text #11

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.3363. Min Act: -0.1700

Data Index: 1183279 (C4 (Web Text))

Max Activating Token Index: 835

Text #12

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.4300. Min Act: -0.1700

Data Index: 739502 (C4 (Web Text))

Max Activating Token Index: 66

Text #13

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.3675. Min Act: -0.1700

Data Index: 743578 (C4 (Web Text))

Max Activating Token Index: 756

Text #14

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.3363. Min Act: -0.1700

Data Index: 1091669 (C4 (Web Text))

Max Activating Token Index: 182

Text #15

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.3363. Min Act: -0.1700

Data Index: 763196 (C4 (Web Text))

Max Activating Token Index: 12

Text #16

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.3363. Min Act: -0.1700

Data Index: 1432205 (Python Code)

Max Activating Token Index: 339

Text #17

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.3363. Min Act: -0.1700

Data Index: 944440 (C4 (Web Text))

Max Activating Token Index: 31

Text #18

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.2738. Min Act: -0.1700

Data Index: 451084 (C4 (Web Text))

Max Activating Token Index: 980

Text #19

Max Range: 5.7425. Min Range: -5.7425

Max Act: 5.3675. Min Act: -0.1700

Data Index: 1187855 (C4 (Web Text))

Max Activating Token Index: 48