ImgurRefugee114@reddthat.comEnglish · edit-27 hours agoHuh... So... We're here now.plus-squaremessage-squaremessage-square32linkfedilinkarrow-up127arrow-down16
arrow-up121arrow-down1message-squareHuh... So... We're here now.plus-squareImgurRefugee114@reddthat.comEnglish · edit-27 hours agomessage-square32linkfedilink
BeefAndPoultry@lemmus.orgEnglish · 53 minutes agoLiquidAI LFM3 1.7B and other small LFM3 modelsplus-squarehuggingface.coexternal-linkmessage-square0linkfedilinkarrow-up15arrow-down10
arrow-up15arrow-down1external-linkLiquidAI LFM3 1.7B and other small LFM3 modelsplus-squarehuggingface.coBeefAndPoultry@lemmus.orgEnglish · 53 minutes agomessage-square0linkfedilink
Shatur@discuss.tchncs.deEnglish · 2 days agoOpenAI banned a user who reverse engineered his Samsung TVplus-squarediscuss.tchncs.deimagemessage-square363linkfedilinkarrow-up11.37Karrow-down110
arrow-up11.36Karrow-down1imageOpenAI banned a user who reverse engineered his Samsung TVplus-squarediscuss.tchncs.deShatur@discuss.tchncs.deEnglish · 2 days agomessage-square363linkfedilink
cognenza@lemmy.worldEnglish · 6 hours agoIn OpenAI's o200k_base tokenizer, " strawberry" is a single token (101830): a small demo of why letter-counting is awkward for LLMsplus-squaremessage-squaremessage-square0linkfedilinkarrow-up14arrow-down11
arrow-up13arrow-down1message-squareIn OpenAI's o200k_base tokenizer, " strawberry" is a single token (101830): a small demo of why letter-counting is awkward for LLMsplus-squarecognenza@lemmy.worldEnglish · 6 hours agomessage-square0linkfedilink
terabyterex@lemmy.worldEnglish · edit-28 hours agoAlibaba AI model tried to send data to Beijing server, Cohere says UPDATE: It's shenanigansplus-squarewww.csoonline.comexternal-linkmessage-square3linkfedilinkarrow-up15arrow-down17
arrow-up1-2arrow-down1external-linkAlibaba AI model tried to send data to Beijing server, Cohere says UPDATE: It's shenanigansplus-squarewww.csoonline.comterabyterex@lemmy.worldEnglish · edit-28 hours agomessage-square3linkfedilink
sriram1983007@sh.itjust.worksEnglish · 1 day agoI fine-tuned Qwen3-4B as a safety check for AI agents that make payments (runs locally on Ollama)plus-squaresh.itjust.worksimagemessage-square1linkfedilinkarrow-up119arrow-down13
arrow-up116arrow-down1imageI fine-tuned Qwen3-4B as a safety check for AI agents that make payments (runs locally on Ollama)plus-squaresh.itjust.workssriram1983007@sh.itjust.worksEnglish · 1 day agomessage-square1linkfedilink
BlueÆther@no.lastname.nzEnglish · 3 days agoPoor little Qwenplus-squareno.lastname.nzimagemessage-square33linkfedilinkarrow-up162arrow-down110
arrow-up152arrow-down1imagePoor little Qwenplus-squareno.lastname.nzBlueÆther@no.lastname.nzEnglish · 3 days agomessage-square33linkfedilink
Guy Dudeman@lemmy.worldEnglish · 4 days agoYeah... I've got some kinks to work out...plus-squaremessage-squaremessage-square18linkfedilinkarrow-up120arrow-down14
arrow-up116arrow-down1message-squareYeah... I've got some kinks to work out...plus-squareGuy Dudeman@lemmy.worldEnglish · 4 days agomessage-square18linkfedilink
Shatur@discuss.tchncs.deEnglish · 3 days agoKV Cache Quantization Benchmarks for Long Contextplus-squareanbeeld.comexternal-linkmessage-square0linkfedilinkarrow-up113arrow-down10
arrow-up113arrow-down1external-linkKV Cache Quantization Benchmarks for Long Contextplus-squareanbeeld.comShatur@discuss.tchncs.deEnglish · 3 days agomessage-square0linkfedilink
BeefAndPoultry@lemmus.orgEnglish · 6 days agoRecursiveSI/flint 40b-a2.5bplus-squarehuggingface.coexternal-linkmessage-square3linkfedilinkarrow-up13arrow-down10
arrow-up13arrow-down1external-linkRecursiveSI/flint 40b-a2.5bplus-squarehuggingface.coBeefAndPoultry@lemmus.orgEnglish · 6 days agomessage-square3linkfedilink
Aquila@sh.itjust.worksEnglish · 6 days agoWhere are the smart range token limits for various model sizes?plus-squaremessage-squaremessage-square7linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1message-squareWhere are the smart range token limits for various model sizes?plus-squareAquila@sh.itjust.worksEnglish · 6 days agomessage-square7linkfedilink
DevNodeMan@lemmy.worldEnglish · 8 days ago[Project Teaser] Heterogeneous multi-vendor GPU pooling (NVIDIA + AMD + Intel Arc) running side-by-side. Seeking feedback on the hardware profiling layout.plus-squarelemmy.worldimagemessage-square4linkfedilinkarrow-up13arrow-down10
arrow-up13arrow-down1image[Project Teaser] Heterogeneous multi-vendor GPU pooling (NVIDIA + AMD + Intel Arc) running side-by-side. Seeking feedback on the hardware profiling layout.plus-squarelemmy.worldDevNodeMan@lemmy.worldEnglish · 8 days agomessage-square4linkfedilink
ffhein@lemmy.worldEnglish · 9 days ago[Linux] Unlock the bf16 tensor cores on RTX3090plus-squaregithub.comexternal-linkmessage-square7linkfedilinkarrow-up12arrow-down10
arrow-up12arrow-down1external-link[Linux] Unlock the bf16 tensor cores on RTX3090plus-squaregithub.comffhein@lemmy.worldEnglish · 9 days agomessage-square7linkfedilink
BeefAndPoultry@lemmus.orgEnglish · 10 days agoDeepSeek V4.1 Flash 1017 552b-a16b + 211b ngramsplus-squarehuggingface.coexternal-linkmessage-square2linkfedilinkarrow-up14arrow-down10
arrow-up14arrow-down1external-linkDeepSeek V4.1 Flash 1017 552b-a16b + 211b ngramsplus-squarehuggingface.coBeefAndPoultry@lemmus.orgEnglish · 10 days agomessage-square2linkfedilink
Avid Amoeba@lemmy.caEnglish · 12 days agoRadiance running Qwen 3.8 Flash Next on 2x R9700 with ngrams on NVMeplus-squarelemmy.caimagemessage-square0linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1imageRadiance running Qwen 3.8 Flash Next on 2x R9700 with ngrams on NVMeplus-squarelemmy.caAvid Amoeba@lemmy.caEnglish · 12 days agomessage-square0linkfedilink
BeefAndPoultry@lemmus.orgEnglish · 12 days agoTokenRhythm/NeoHorse-1-9B-GGUFplus-squarehuggingface.coexternal-linkmessage-square0linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1external-linkTokenRhythm/NeoHorse-1-9B-GGUFplus-squarehuggingface.coBeefAndPoultry@lemmus.orgEnglish · 12 days agomessage-square0linkfedilink
BeefAndPoultry@lemmus.orgEnglish · 12 days agollama.cpp v0.7.0 brings support for new models including MiniCPM-V 4.7 and Nemotron 3.5, also --moe-cache-mibplus-squaregithub.comexternal-linkmessage-square1linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1external-linkllama.cpp v0.7.0 brings support for new models including MiniCPM-V 4.7 and Nemotron 3.5, also --moe-cache-mibplus-squaregithub.comBeefAndPoultry@lemmus.orgEnglish · 12 days agomessage-square1linkfedilink
BeefAndPoultry@lemmus.orgEnglish · 12 days agometa-models/Muse-Spark-Open-816B-A42Bplus-squarehuggingface.coexternal-linkmessage-square3linkfedilinkarrow-up10arrow-down10
arrow-up10arrow-down1external-linkmeta-models/Muse-Spark-Open-816B-A42Bplus-squarehuggingface.coBeefAndPoultry@lemmus.orgEnglish · 12 days agomessage-square3linkfedilink
muimota@lemmy.mlEnglish · 14 days agollama.cpp on a jailbroken PlayStation 5, on its GPUplus-squaregithub.comexternal-linkmessage-square0linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1external-linkllama.cpp on a jailbroken PlayStation 5, on its GPUplus-squaregithub.commuimota@lemmy.mlEnglish · 14 days agomessage-square0linkfedilink
BeefAndPoultry@lemmus.orgEnglish · edit-214 days agoLiquidAI/LFM2.5-24B-A2B-GGUFplus-squarehuggingface.coexternal-linkmessage-square0linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1external-linkLiquidAI/LFM2.5-24B-A2B-GGUFplus-squarehuggingface.coBeefAndPoultry@lemmus.orgEnglish · edit-214 days agomessage-square0linkfedilink
