Model: GELU Model: 3 Layers, 2048 Neurons per Layer
Dataset: 80% C4 (Web Text) and 20% Python Code
Neuron 1997 in Layer 1
Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')
Text #0
Max Range: 4.2009. Min Range: -4.2009
Max Act: 4.2009. Min Act: -0.1700
Data Index: 508 (C4 (Web Text))
Max Activating Token Index: 221
Click toggle to see full text
Truncated
Full Text #0
Text #1
Max Range: 4.2009. Min Range: -4.2009
Max Act: 4.0758. Min Act: -0.1700
Data Index: 319818 (C4 (Web Text))
Max Activating Token Index: 278
Click toggle to see full text
Truncated
Full Text #1
Text #2
Max Range: 4.2009. Min Range: -4.2009
Max Act: 4.0446. Min Act: -0.1700
Data Index: 238515 (C4 (Web Text))
Max Activating Token Index: 378
Click toggle to see full text
Truncated
Full Text #2
Text #3
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.9508. Min Act: -0.1700
Data Index: 950247 (C4 (Web Text))
Max Activating Token Index: 137
Click toggle to see full text
Truncated
Full Text #3
Text #4
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.8882. Min Act: -0.1700
Data Index: 901028 (C4 (Web Text))
Max Activating Token Index: 161
Click toggle to see full text
Truncated
Full Text #4
Text #5
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.8882. Min Act: -0.1700
Data Index: 871049 (C4 (Web Text))
Max Activating Token Index: 199
Click toggle to see full text
Truncated
Full Text #5
Text #6
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.8882. Min Act: -0.1700
Data Index: 418719 (C4 (Web Text))
Max Activating Token Index: 427
Click toggle to see full text
Truncated
Full Text #6
Text #7
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.8882. Min Act: -0.1700
Data Index: 565209 (C4 (Web Text))
Max Activating Token Index: 282
Click toggle to see full text
Truncated
Full Text #7
Text #8
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.8569. Min Act: -0.1700
Data Index: 1065510 (C4 (Web Text))
Max Activating Token Index: 133
Click toggle to see full text
Truncated
Full Text #8
Text #9
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.8569. Min Act: -0.1700
Data Index: 1229737 (C4 (Web Text))
Max Activating Token Index: 219
Click toggle to see full text
Truncated
Full Text #9
Text #10
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.8257. Min Act: -0.1700
Data Index: 1070804 (C4 (Web Text))
Max Activating Token Index: 196
Click toggle to see full text
Truncated
Full Text #10
Text #11
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.7944. Min Act: -0.1700
Data Index: 1472299 (Python Code)
Max Activating Token Index: 141
Click toggle to see full text
Truncated
Full Text #11
Text #12
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.7944. Min Act: -0.1700
Data Index: 1015738 (C4 (Web Text))
Max Activating Token Index: 293
Click toggle to see full text
Truncated
Full Text #12
Text #13
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.7944. Min Act: -0.1700
Data Index: 1085027 (C4 (Web Text))
Max Activating Token Index: 235
Click toggle to see full text
Truncated
Full Text #13
Text #14
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.7318. Min Act: -0.1700
Data Index: 545266 (C4 (Web Text))
Max Activating Token Index: 200
Click toggle to see full text
Truncated
Full Text #14
Text #15
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.7631. Min Act: -0.1700
Data Index: 26401 (C4 (Web Text))
Max Activating Token Index: 159
Click toggle to see full text
Truncated
Full Text #15
Text #16
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.7318. Min Act: -0.1700
Data Index: 652888 (C4 (Web Text))
Max Activating Token Index: 208
Click toggle to see full text
Truncated
Full Text #16
Text #17
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.7318. Min Act: -0.1700
Data Index: 355445 (C4 (Web Text))
Max Activating Token Index: 430
Click toggle to see full text
Truncated
Full Text #17
Text #18
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.7631. Min Act: -0.1699
Data Index: 1359973 (Python Code)
Max Activating Token Index: 216
Click toggle to see full text
Truncated
Full Text #18
Text #19
Max Range: 4.2009. Min Range: -4.2009
Max Act: 3.6692. Min Act: -0.1700
Data Index: 110069 (C4 (Web Text))
Max Activating Token Index: 320
Click toggle to see full text
Truncated
Full Text #19