21–30 of 125 posts
It just baffles me that this even works at all. Doesn't it just become a token soup? Is attention that precise that a second token can tell its the second token just because it learns to accumulate something in the embedding space without any sort of inductive bias?
"Interestingly, Kimi K3 got rid of all RoPE layers and uses NoPE (No Positional Embeddings) everywhere instead." It just baffles me that this even works at all. Doesn't it just become a token soup? Is attention that precise that a second token can tell its the second token just because it learns to accumulate something in the embedding space without any sort of inductive bias?
Sabastian Raschka is one of the great LLM researchers/authors. I highly recommend his substack
"Interestingly, Kimi K3 got rid of all RoPE layers and uses NoPE (No Positional Embeddings) everywhere instead." It just baffles me that this even works at all. Doesn't it just become a token soup? Is attention that precise that a second token can tell its the second token just because it learns to accumulate something in the embedding space without any sort of inductive bias?
And yes accumulation is a good intuition for what's going on. You could imagine a part of the attention head that just kept writing to the same part of the residual stream causing that to keep accumulating (simply via attention summation) as more input tokens come in thereby functioning as a kind of index without the need for any positional encoding.
"Interestingly, Kimi K3 got rid of all RoPE layers and uses NoPE (No Positional Embeddings) everywhere instead." It just baffles me that this even works at all. Doesn't it just become a token soup? Is attention that precise that a second token can tell its the second token just because it learns to accumulate something in the embedding space without any sort of inductive bias?