cm0002@lemmings.worldEnglish · 13 hours agoCutting LLM Memory by 84%: A Deep Dive into Fused Kernelsplus-squaretowardsdatascience.comexternal-linkmessage-square0linkfedilinkarrow-up15arrow-down12
arrow-up13arrow-down1external-linkCutting LLM Memory by 84%: A Deep Dive into Fused Kernelsplus-squaretowardsdatascience.comcm0002@lemmings.worldEnglish · 13 hours agomessage-square0linkfedilink
cm0002@lemmings.worldEnglish · 23 hours agoShrinking AI memory boosts accuracyplus-squarewww.ed.ac.ukexternal-linkmessage-square0linkfedilinkarrow-up15arrow-down10
arrow-up15arrow-down1external-linkShrinking AI memory boosts accuracyplus-squarewww.ed.ac.ukcm0002@lemmings.worldEnglish · 23 hours agomessage-square0linkfedilink
cm0002@lemmy.cafeEnglish · 3 days agoMistral Small Creative beats Claude Opus 4.5 at explaining transformers — 50x cheaper, higher scoresplus-squaresubstack.comexternal-linkmessage-square2linkfedilinkarrow-up14arrow-down12
arrow-up12arrow-down1external-linkMistral Small Creative beats Claude Opus 4.5 at explaining transformers — 50x cheaper, higher scoresplus-squaresubstack.comcm0002@lemmy.cafeEnglish · 3 days agomessage-square2linkfedilink
atro_city@fedia.io · 5 days agoTurn Microsoft into "Microslop" everywhere with this new browser extension — CEO Satya Nadella discourages the term, but it's having the opposite effectplus-squarefedia.ioimagemessage-square1linkfedilinkarrow-up146arrow-down10
arrow-up146arrow-down1imageTurn Microsoft into "Microslop" everywhere with this new browser extension — CEO Satya Nadella discourages the term, but it's having the opposite effectplus-squarefedia.ioatro_city@fedia.io · 5 days agomessage-square1linkfedilink
cm0002@lemmy.cafeEnglish · 3 days agoDeepSeek just published a paper on conditional memory via scalable lookupplus-squaregithub.comexternal-linkmessage-square0linkfedilinkarrow-up14arrow-down12
arrow-up12arrow-down1external-linkDeepSeek just published a paper on conditional memory via scalable lookupplus-squaregithub.comcm0002@lemmy.cafeEnglish · 3 days agomessage-square0linkfedilink
codeinaboxEnglish · 4 days agoWeird Generalization and Inductive Backdoors: New Ways to Corrupt LLMsplus-squarearxiv.orgexternal-linkmessage-square0linkfedilinkarrow-up14arrow-down10
arrow-up14arrow-down1external-linkWeird Generalization and Inductive Backdoors: New Ways to Corrupt LLMsplus-squarearxiv.orgcodeinaboxEnglish · 4 days agomessage-square0linkfedilink
cm0002@suppo.fiEnglish · 8 days agoJensen Huang Is Begging You to Stop Being So Negative About AIplus-squaregizmodo.comexternal-linkmessage-square12linkfedilinkarrow-up123arrow-down14
arrow-up119arrow-down1external-linkJensen Huang Is Begging You to Stop Being So Negative About AIplus-squaregizmodo.comcm0002@suppo.fiEnglish · 8 days agomessage-square12linkfedilink
codeinaboxEnglish · 4 days agoWill Your AI Teammate Bring Bagels to Standup?plus-squareredmonk.comexternal-linkmessage-square0linkfedilinkarrow-up12arrow-down11
arrow-up11arrow-down1external-linkWill Your AI Teammate Bring Bagels to Standup?plus-squareredmonk.comcodeinaboxEnglish · 4 days agomessage-square0linkfedilink
cm0002@infosec.pubEnglish · 7 days agoDroPE: Extending the Context of Pretrained LLMs by Dropping their Positional Embeddingsplus-squarepub.sakana.aiexternal-linkmessage-square0linkfedilinkarrow-up13arrow-down11
arrow-up12arrow-down1external-linkDroPE: Extending the Context of Pretrained LLMs by Dropping their Positional Embeddingsplus-squarepub.sakana.aicm0002@infosec.pubEnglish · 7 days agomessage-square0linkfedilink
cm0002@toast.oooEnglish · 11 days agoChatGPT May Be Eroding Critical Thinking Skills, According to a New MIT Studyplus-squaretime.comexternal-linkmessage-square1linkfedilinkarrow-up122arrow-down11
arrow-up121arrow-down1external-linkChatGPT May Be Eroding Critical Thinking Skills, According to a New MIT Studyplus-squaretime.comcm0002@toast.oooEnglish · 11 days agomessage-square1linkfedilink
cm0002@lemdro.idEnglish · 17 days agoI don't care how well your "AI" works - fiona fokusplus-squarefokus.coolexternal-linkmessage-square0linkfedilinkarrow-up13arrow-down11
arrow-up12arrow-down1external-linkI don't care how well your "AI" works - fiona fokusplus-squarefokus.coolcm0002@lemdro.idEnglish · 17 days agomessage-square0linkfedilink
cm0002@no.lastname.nzEnglish · 21 days agoA Convolutional Neural Network implemented entirely from scratch in x86-64 assembly using AVX-512, performing cat vs dog image classification without any ML frameworks or libraries.plus-squaregithub.comexternal-linkmessage-square0linkfedilinkarrow-up110arrow-down11
arrow-up19arrow-down1external-linkA Convolutional Neural Network implemented entirely from scratch in x86-64 assembly using AVX-512, performing cat vs dog image classification without any ML frameworks or libraries.plus-squaregithub.comcm0002@no.lastname.nzEnglish · 21 days agomessage-square0linkfedilink
chasteinsectEnglish · 21 days agoAI is hallucinating its way into research, and that’s not even where the problem startsplus-squarethelibre.newsexternal-linkmessage-square1linkfedilinkarrow-up113arrow-down10
arrow-up113arrow-down1external-linkAI is hallucinating its way into research, and that’s not even where the problem startsplus-squarethelibre.newschasteinsectEnglish · 21 days agomessage-square1linkfedilink
codeinaboxEnglish · 22 days agoThe Gorman Paradox: Where Are All The AI-Generated Apps?plus-squarecodemanship.wordpress.comexternal-linkmessage-square0linkfedilinkarrow-up110arrow-down10
arrow-up110arrow-down1external-linkThe Gorman Paradox: Where Are All The AI-Generated Apps?plus-squarecodemanship.wordpress.comcodeinaboxEnglish · 22 days agomessage-square0linkfedilink
cm0002@lemy.lolEnglish · 22 days agoTencent just released WeDLM 8B, it's a diffusion language model that runs 3-6× faster than vLLM-optimized Qwen3-8B on math reasoning tasks.plus-squarehuggingface.coexternal-linkmessage-square0linkfedilinkarrow-up14arrow-down12
arrow-up12arrow-down1external-linkTencent just released WeDLM 8B, it's a diffusion language model that runs 3-6× faster than vLLM-optimized Qwen3-8B on math reasoning tasks.plus-squarehuggingface.cocm0002@lemy.lolEnglish · 22 days agomessage-square0linkfedilink
onlinepersonaEnglish · 23 days agoThe AI-collapse pre-mortem - Bert Hubert's writingsplus-squareberthub.euexternal-linkmessage-square0linkfedilinkarrow-up118arrow-down11
arrow-up117arrow-down1external-linkThe AI-collapse pre-mortem - Bert Hubert's writingsplus-squareberthub.euonlinepersonaEnglish · 23 days agomessage-square0linkfedilink
cm0002@lemmy.cafeEnglish · 25 days agoThe Computer Chronicles - Artificial Intelligence (1984)plus-squarewww.youtube.comexternal-linkmessage-square0linkfedilinkarrow-up13arrow-down10
arrow-up13arrow-down1external-linkThe Computer Chronicles - Artificial Intelligence (1984)plus-squarewww.youtube.comcm0002@lemmy.cafeEnglish · 25 days agomessage-square0linkfedilink
cm0002@toast.oooEnglish · 25 days agoHow to Deal With AI Restrictions When Upscaling Imagesplus-squareivanca.github.ioexternal-linkmessage-square0linkfedilinkarrow-up15arrow-down12
arrow-up13arrow-down1external-linkHow to Deal With AI Restrictions When Upscaling Imagesplus-squareivanca.github.iocm0002@toast.oooEnglish · 25 days agomessage-square0linkfedilink
cm0002@mander.xyzEnglish · 28 days agoI'm running a 120B local LLM on 24GB of VRAM, and now it powers my smart homeplus-squarewww.xda-developers.comexternal-linkmessage-square2linkfedilinkarrow-up15arrow-down15
arrow-up10arrow-down1external-linkI'm running a 120B local LLM on 24GB of VRAM, and now it powers my smart homeplus-squarewww.xda-developers.comcm0002@mander.xyzEnglish · 28 days agomessage-square2linkfedilink
monica_b1998@lemmy.worldEnglish · 28 days agoGuided learning lets “untrainable” neural networks realize their potentialplus-squarenews.mit.eduexternal-linkmessage-square0linkfedilinkarrow-up16arrow-down11
arrow-up15arrow-down1external-linkGuided learning lets “untrainable” neural networks realize their potentialplus-squarenews.mit.edumonica_b1998@lemmy.worldEnglish · 28 days agomessage-square0linkfedilink
