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

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

Neuron 866 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: 5.7764. Min Range: -5.7764

Max Act: 5.7764. Min Act: -0.1700

Data Index: 391798 (C4 (Web Text))

Max Activating Token Index: 178

Text #1

Max Range: 5.7764. Min Range: -5.7764

Max Act: 5.3701. Min Act: -0.1700

Data Index: 324679 (C4 (Web Text))

Max Activating Token Index: 554

Text #2

Max Range: 5.7764. Min Range: -5.7764

Max Act: 5.2139. Min Act: -0.1700

Data Index: 763584 (C4 (Web Text))

Max Activating Token Index: 542

Text #3

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.9326. Min Act: -0.1700

Data Index: 188739 (C4 (Web Text))

Max Activating Token Index: 126

Text #4

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.9951. Min Act: -0.1700

Data Index: 883158 (C4 (Web Text))

Max Activating Token Index: 681

Text #5

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.9639. Min Act: -0.1700

Data Index: 1376491 (Python Code)

Max Activating Token Index: 975

Text #6

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.9639. Min Act: -0.1699

Data Index: 325538 (C4 (Web Text))

Max Activating Token Index: 336

Text #7

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.9014. Min Act: -0.1700

Data Index: 306137 (C4 (Web Text))

Max Activating Token Index: 299

Text #8

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.8701. Min Act: -0.1700

Data Index: 1364407 (Python Code)

Max Activating Token Index: 932

Text #9

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.9326. Min Act: -0.1700

Data Index: 1461210 (Python Code)

Max Activating Token Index: 179

Text #10

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.8701. Min Act: -0.1700

Data Index: 82832 (C4 (Web Text))

Max Activating Token Index: 567

Text #11

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.8701. Min Act: -0.1700

Data Index: 1437181 (Python Code)

Max Activating Token Index: 630

Text #12

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.8389. Min Act: -0.1700

Data Index: 482466 (C4 (Web Text))

Max Activating Token Index: 291

Text #13

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.8076. Min Act: -0.1700

Data Index: 923719 (C4 (Web Text))

Max Activating Token Index: 222

Text #14

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.8076. Min Act: -0.1700

Data Index: 1577617 (Python Code)

Max Activating Token Index: 669

Text #15

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.7764. Min Act: -0.1700

Data Index: 635109 (C4 (Web Text))

Max Activating Token Index: 516

Text #16

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.7451. Min Act: -0.1700

Data Index: 1530481 (Python Code)

Max Activating Token Index: 517

Text #17

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.6826. Min Act: -0.1700

Data Index: 1497808 (Python Code)

Max Activating Token Index: 968

Text #18

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.7451. Min Act: -0.1700

Data Index: 1646025 (Python Code)

Max Activating Token Index: 722

Text #19

Max Range: 5.7764. Min Range: -5.7764

Max Act: 4.7451. Min Act: -0.1700

Data Index: 1542363 (Python Code)

Max Activating Token Index: 490