Fast KV Compaction via Attention Matching

Fast KV Compaction via Attention Matching(arxiv.org)

73 points by cbracketdash 87 days ago | 15 comments

WarmWash 87 days ago |

Considering the insanity of the AI arms race going on now, and the incredible sums of money be thrown at any slight advantage, is there any reason to believe that any meaningful AI breakthrough would be openly published for anyone to leverage?

542458 86 days ago | |

These folks are MIT, so citations are valuable to them. Citations convert into prestige, academic career progression, or a favorable exit from academia into industry.

Also, I don't see why you couldn't patent this if you wanted to monetize it.

BetaDeltaAlpha 86 days ago | | |

> Also, I don't see why you couldn't patent this if you wanted to monetize it.

We all just saw the prior art published for the public. That will preclude patenting this work. Further reduction to practice is required.

(I am not a lawyer).

jph00 86 days ago | |

Yes there is. Lots of researchers are more interested in making a contribution to societal flourishing than in making incredible sums of money. That’s why there’s still lots of top AI researchers in academia.

abeppu 86 days ago | |

I do sometimes wonder -- if the transformers paper wasn't published, what would the industry be like? Would the same ideas have been put together in almost the same way weeks or months later somewhere else?

mikodin 86 days ago | |

I would say yes.

The reality is that the money being thrown = the time of humans. I guess compute as well, but in terms of people doing innovation - openly published things are the same thing, minus the money.

cma 86 days ago | |

The inventor's grace period under first to file changes still gives them/their university a year to file if they publish openly.

gdiamos 86 days ago | |

I know the frontier “labs” are holding back publications.

I don’t think it will last among researchers who think beyond production LLMs

cadamsdotcom 87 days ago |

Superficially it sounds like this could create a bit more of a move toward doing compaction on some continuous basis, or compacting in batches once you hit the context limit, rather than starting fresh with a summary and system prompt..

Feels like high fidelity, fast compaction could be a path to “solving” long context.

cs702 86 days ago |

This looks promising. I've added it to my reading list.

speedping 87 days ago |

This is big for long-horizon tasks

esafak 87 days ago |

None of the compaction accuracies look impressive.

yorwba 87 days ago | |

I think matching or exceeding the original cache at 20% compacted size is fairly impressive.

esafak 86 days ago | | |

The original cache had 70% accuracy, and the alternatives were only worse.