Hubi@feddit.org to Technology@lemmy.worldEnglish · 1 day ago“They Never Even Unpacked the Racks”: Police Raid on Nvidia Partner Megaspeed Finds Empty Server Frames and Missing GPUs Amid Allegations Chips Were Flown to China During Retail Shortagesrespawnfirst.comexternal-linkmessage-square150linkfedilinkarrow-up1830arrow-down111
arrow-up1819arrow-down1external-link“They Never Even Unpacked the Racks”: Police Raid on Nvidia Partner Megaspeed Finds Empty Server Frames and Missing GPUs Amid Allegations Chips Were Flown to China During Retail Shortagesrespawnfirst.comHubi@feddit.org to Technology@lemmy.worldEnglish · 1 day agomessage-square150linkfedilink
minus-squareLooseLemons@lemmy.worldlinkfedilinkEnglisharrow-up2arrow-down10·1 day agoYou’re not sure how at home training is relevant when we’re discussing the output of LLMs? Or home sepecifc stuff?
minus-squarebrucethemoose@lemmy.worldlinkfedilinkEnglisharrow-up4·1 day agoHome inference. And that’s a very broad statement. But nothing we can do locally will get an LLM to intuit or reveal its own dataset. We can get it to output Claude traces it was trained on. I am not trying to be pedantic, but the semantics/phrasing there is very important.
minus-squareLooseLemons@lemmy.worldlinkfedilinkEnglisharrow-up1arrow-down7·1 day agoNot byte for byte, but one can prove an llm was trained on data by asking specific questions. It’s why we’re having all of the copyright lawsuits.
You’re not sure how at home training is relevant when we’re discussing the output of LLMs? Or home sepecifc stuff?
Home inference.
And that’s a very broad statement. But nothing we can do locally will get an LLM to intuit or reveal its own dataset.
We can get it to output Claude traces it was trained on. I am not trying to be pedantic, but the semantics/phrasing there is very important.
Not byte for byte, but one can prove an llm was trained on data by asking specific questions. It’s why we’re having all of the copyright lawsuits.