Re:Kolmogorov-Arnold Networks
by kolinko on Wednesday May 01, 2024 @03:47PM
(#40228431)
I may be wrong but with midern llms biases aren’t really used any more.
[ Parent
| on HN ]
Re:Kolmogorov-Arnold Networks
by tripplyons on Wednesday May 01, 2024 @05:51PM
(#40229971)
From what I remember, larger LLMs like PaLM don't use biases for training stability, but smaller ones tend to still use them.