Abed

Founder of Tech Guard. Building custom software and on-premise AI systems for businesses; client-owned infrastructure. Own your AI, own your data.

[Ash Hart @ashxhart]

"TensorFold did it ~3x faster" 🫶🏼

I am still plowing through the PR's and issues, but more performance will be unlocked soon.

[Abed @TechGuardAI]

@TheAhmadOsman

[Abed @TechGuardAI]

@TheAhmadOsman

[Abed @TechGuardAI]

@u1tra_instinct Now if we can figure out the concurrency with tensorfold for the 2x sparks ⚡️

[Abed @TechGuardAI]

@ayayalar Glm 5.3 flash

[Abed @TechGuardAI]

@Beamsters1 Where’s the github repo for it my friend 😁

[Abed @TechGuardAI]

@ashxhart hugging face link or git repo please 😁

[Abed @TechGuardAI]

@VadimStrizheus Your data

[Abed @TechGuardAI]

@TheAhmadOsman @onusoz Unless 2x cluster cx7 at 200Gb/s

[Abed @TechGuardAI]

@imost Glm better locally

[Abed @TechGuardAI]

@ksuniri Hope you get it, we’re running one of these clusters TP2 DGX for GLM and it’s the real deal.

[Abed @TechGuardAI]

@ElectBilzerian @QueenCrystal22 Bring us in and we’ll help you build your own AI setup locally, your AI, data and tools stay within your network. 🫡

[Abed @TechGuardAI]

@chadaveau @MiaAI_lab 💯

[Abed @TechGuardAI]

@Tech2Wild Idk bro I feel like glm flash quality still wins

[mr_r0b0t @mr_r0b0t]

@NVIDIAAI gifting us another NVFP4!

nvidia/GLM-5.3-Flash-NVFP4

Disk size and GPU memory requirements reduced by approximately 3.33x vs BF16 😍
https://t.co/vQWCus1m8J

[Abed @TechGuardAI]

@MichaelHutu Glm 5.3 flash better than both

[Abed @TechGuardAI]

@p3n73st3r @ashishgupta1103 @HuggingModels @grok 😆😆

[Abed @TechGuardAI]

@DouglasYaoDY

[Tech2Wild @Tech2Wild]

WAKE UP ! DFLASH IS LIVE ! https://t.co/huq5j8OLTC

[Abed @TechGuardAI]

@0xWhiteMage Don’t worry we will make this fit on a single spark just like we did with DSV4 Flash 0731