Model: SoLU Model: 10 Layers, 5120 Neurons per Layer
Dataset: 80% C4 (Web Text) and 20% Python Code
Neuron 4072 in Layer 3
Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')
Text #0
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7880. Min Act: -0.0001
Data Index: 1553930 (Python Code)
Max Activating Token Index: 564
Click toggle to see full text
Truncated
Full Text #0
Text #1
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.8046. Min Act: -0.0001
Data Index: 1652097 (Python Code)
Max Activating Token Index: 453
Click toggle to see full text
Truncated
Full Text #1
Text #2
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7847. Min Act: -0.0001
Data Index: 1481170 (Python Code)
Max Activating Token Index: 430
Click toggle to see full text
Truncated
Full Text #2
Text #3
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7199. Min Act: -0.0001
Data Index: 496729 (C4 (Web Text))
Max Activating Token Index: 766
Click toggle to see full text
Truncated
Full Text #3
Text #4
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7696. Min Act: -0.0001
Data Index: 1382514 (Python Code)
Max Activating Token Index: 476
Click toggle to see full text
Truncated
Full Text #4
Text #5
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7879. Min Act: -0.0001
Data Index: 1370649 (Python Code)
Max Activating Token Index: 256
Click toggle to see full text
Truncated
Full Text #5
Text #6
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7827. Min Act: -0.0001
Data Index: 1576495 (Python Code)
Max Activating Token Index: 468
Click toggle to see full text
Truncated
Full Text #6
Text #7
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7235. Min Act: -0.0001
Data Index: 517777 (C4 (Web Text))
Max Activating Token Index: 707
Click toggle to see full text
Truncated
Full Text #7
Text #8
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.6414. Min Act: -0.0001
Data Index: 42544 (C4 (Web Text))
Max Activating Token Index: 748
Click toggle to see full text
Truncated
Full Text #8
Text #9
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7447. Min Act: -0.0001
Data Index: 1525240 (Python Code)
Max Activating Token Index: 684
Click toggle to see full text
Truncated
Full Text #9
Text #10
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.6705. Min Act: -0.0001
Data Index: 500763 (C4 (Web Text))
Max Activating Token Index: 535
Click toggle to see full text
Truncated
Full Text #10
Text #11
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7080. Min Act: -0.0001
Data Index: 1630102 (Python Code)
Max Activating Token Index: 726
Click toggle to see full text
Truncated
Full Text #11
Text #12
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7424. Min Act: -0.0001
Data Index: 373669 (C4 (Web Text))
Max Activating Token Index: 17
Click toggle to see full text
Truncated
Full Text #12
Text #13
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7239. Min Act: -0.0001
Data Index: 1368090 (Python Code)
Max Activating Token Index: 257
Click toggle to see full text
Truncated
Full Text #13
Text #14
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7163. Min Act: -0.0001
Data Index: 1560313 (Python Code)
Max Activating Token Index: 832
Click toggle to see full text
Truncated
Full Text #14
Text #15
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.7618. Min Act: -0.0001
Data Index: 491492 (C4 (Web Text))
Max Activating Token Index: 370
Click toggle to see full text
Truncated
Full Text #15
Text #16
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.6691. Min Act: -0.0001
Data Index: 252092 (C4 (Web Text))
Max Activating Token Index: 529
Click toggle to see full text
Truncated
Full Text #16
Text #17
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.6696. Min Act: -0.0001
Data Index: 1656176 (Python Code)
Max Activating Token Index: 497
Click toggle to see full text
Truncated
Full Text #17
Text #18
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.6682. Min Act: -0.0001
Data Index: 1181359 (C4 (Web Text))
Max Activating Token Index: 366
Click toggle to see full text
Truncated
Full Text #18
Text #19
Max Range: 0.8046. Min Range: -0.8046
Max Act: 0.6703. Min Act: -0.0001
Data Index: 1501676 (Python Code)
Max Activating Token Index: 704
Click toggle to see full text
Truncated
Full Text #19