What we talk about when we talk about system designmaheshba.bitbucket.io259 points·by scv119·3 anni fa·35 comments
scv119·3 anni fa·discussThe tweet is referring to a paper that fine tunes Chinese dataset on english base model. I'm not surprised with LoRA's poor result in this setup.
scv119·3 anni fa·discussI believe you can slightly change the flash attention kernel to implement the same kernel of this page attention, since both of them work on the key/value cache at block level.
scv119·3 anni fa·discussPretty cool stuff and the results are amazing. Hoping we will see virtual memory get standardized in pytorch or cuda.