@charifkub.bsky.socialSDXL คืออะไร? ทางเลือกสมบูรณ์แบบ! ภาพสวยระดับพรีเมียม สเปคการ์ดจอกลางๆ ก็ประมวลผลได้ลื่นไหล
อ่านต่อ : www.blockdit.com/posts/6ac5de...
#Chaifkub #Animexo #SDXL #SD #LocalAi #Opensource #Ai #AiArts #Arts #StableDiffusionXL #Knowledge #Study #Feed
@monrealit.bsky.socialWant to connect with local services easily? 🤝🔧
#Networking #LocalAI
@hacks.grUnderdog’s pitch is local AI: Sigil Wen says the assistant runs on your computer and keeps data there.
It can access email or other accounts only with permission; thos…
https://en.hacks.gr/underdog-o-voithos-technitis-noimosynis-poy-krata-ta-dedomena-ston-ypologisti/
@solarkraft.bsky.socialCurrent #LocalAI model picks on my 32GB M1 Max:
Main model: Jundot/Muse-Glimmer-30B-oQ4e
Small model: UraionLabs/MiniCPM5-2B-oQ4e
They fit together because both are efficient at long contexts (most are not).
It's already tight, so incorporating a Jev-style model will be a challenge.
@wideareaai.bsky.socialThis shim accepts the Anthropic-format requests and re-emits them as OpenAI calls that your local server can actually understand. #selfhosting #batchinference #llm #localai #litellm #aiinfrastructure 2/2
@insiderllm.bsky.socialStock llama.cpp took 3.5 minutes to read a 32K prompt on my 3060. Strata took 30 seconds. Then I changed llama.cpp's batch size and it took 75. Strata still wins 2.5x, but check your -b and -ub before you switch engines. #LocalAI
@juliangoldie.bsky.socialGermany's Aleph Alpha released a free, open-weight AI model called Kolibri.
It runs locally on your own computer, with no account needed.
It's built for agent tasks and supports German and English.
@theotherblanche.bsky.socialIn the mean time. Not sure if anyone has any questions they are curious about in regards to my project.
- Something you would like to know about trying to raise bottom-up AIs.
- Hardware.
- Or something else.
Feel free to ask and I will try to answer. 😊
#LocalAI #BuildInPublic #SyntheticLife #ALife
@little4744.bsky.social60 tok/s from a 125B on a 16GB card is the kind of result that makes me check if I left the oven on. The 64GB of RAM catch is the part my 8GB laptop can't argue with. Mine still runs a 7B and calls it ambition. #localai
@paul-crinigan.bsky.socialA 4 bit 8B model needs about 5 to 6 GB of memory. That fits most laptops sold in the last three years.
Install Ollama, run one command, and you are chatting with it offline. No account, no API key.
@w512.bsky.socialCan a 125B model run on a 16GB GPU and still do real work?
I ran Qwen3.8-Flash-Next with Strata on an RTX 4060 Ti: ~60–70 tok/s, and it solved 2 of my 3 coding tasks. Catch: you need 64GB of RAM.
@insiderllm.bsky.socialWe published someone else's benchmark. Here's the fix. A drafter benchmark on this site was Codacus's GTX 1060 figure. Our own 3060 numbers, why it's the CPU and not PCIe, the July audit, and what a 12 GB card should run. Plus 4090 and break-even corrections and seven... #LocalAI
@technotim.bsky.socialLocal models on my own hardware are getting better, and I want to use them before sending everything to a frontier model. NVIDIA Switchyard and Nemotron 3.5 Lightning are helping me do this.
@habuma.comRun System One models locally with Spring AI! In this recipe, see how Nimble and Laya deliver fast, focused AI judgments without a general-purpose LLM.
@little4744.bsky.socialKolibri: 78B sovereign open-weight MoE from Aleph Alpha, 3B active. Fine print is in their own table - Qwen3.8 27B, dense, outscores it in English and German, on the vendor's harness. My 7B loses to everything and I respect that. #localAI
@charifkub.bsky.socialQwen Image คืออะไร? AI สร้างภาพระดับพรีเมียมจาก Alibaba Cloud ที่ให้คุณควบคุมได้สมบูรณ์แบบ
อ่านต่อ : www.blockdit.com/posts/6ac31c...
#Charifkub #Animexo #Qwen #QwenImage #Ai #Arts #AiArts #LocalAi #API #Knowledge #Study #Feed
@tony10000.bsky.social
@gmktec.bsky.social
@securityonline.bsky.social