keys ๐งช
@u1tra_instinct
#Bitcoin# ๐๐ . ๐งช๐งช๐งช (always testing and get into trouble).
755 Following 5.4K Followers
๐ฝ๐ฝ๐ฝ :::Breaking: @ashxhart Tensorfold has penetrated into Windows11 Arm64.
:: First Benchmark of RTX-Spark performance Fine tune of escaping the windows virtual linux WSL into native windows:::
here is the result using Cyber Frost 3.8 (single spark recipe) by @Blackfrost_AI @ViC305 . just like Neil Armstrong landed on the moon, We have landed in Windows 11.
Desktop performance on a laptop via RTX Spark and Tensorfold is insane.. all running locally, no cloud.
Show more
10% is old news.. it's fucking 28% now.. LFG!!!!!>.
letโs beat HiggsField with a true open source for the @comfyUI open source competition, we can do better and we stand with the community , local , open source, world wide for everyone.!!!!! LFG
the Ablit for MiaLab Recipe just got a speed boost please update new F16 ablit vs F32 head now closer to stock.
Two ablit recipe now in the same repo
LibertAI DeepSeek v4.1 flash 1017 - exl3 - 3.5 BPW now abliterated would support @Tech2Wild TensorFold Recipe please use the same HugginFace Report : link to the HF repo: ๐๐งต
Show more
@ashxhart Amazing work, @u1tra_instinct - and thank you for open sourcing it! I've got something fun built on top of it and TensorFold for @OmarchyMac that I should be able to announce shortly, just waiting on a @huggingface upload to complete. ๐
Show more
@u1tra_instinct Oh yes oh yeah oh yeah!!!!Wooohoooo
๐๐ช open source power
Credit where it's due: MiniMax H3 is the model. The engine is @ashxhart's TensorFold, with H3 support by @u1tra_instinct and the MLX port from @pipenetwork and @AIBizarrothe.
My part: the Linux on Apple Silicon port. The int8 weights are on Hugging Face:
Show more
@u1tra_instinct Beautiful!
Bro your killing it. Thank you for all the hard work
LFG I can see this be part of the omarchy as default tensorfold studio app for the agents and omarchy Linus OS
Can inference directly from it as a feature built into the OS
Credit where it's due: MiniMax H3 is the model. The engine is @ashxhart's TensorFold, with H3 support by @u1tra_instinct and the MLX port from @pipenetwork and @AIBizarrothe.
My part: the Linux on Apple Silicon port. The int8 weights are on Hugging Face:
Show more
TTFT is non-existent for fresh sessions in TensorFold.
Mixing precomputed KV Cache with TensorFold delivers @Alibaba_Qwen 3.8 Flash Next... on a SINGLE DGX Spark - 0.08s TTFT and over 50tps - all thanks to @ashxhart โฅ๏ธ
Preview on the DGX-Spark tensorfold studio
;:::: stay tuned::::: needs only One Dgx-spark
Insane work by Keys building on top of TensorFold to bring video generation to Apple ๐๐ป๐
Thanks, Mia, for believing in TensorFold and pouring countless hours into tuning your recipes to run even quicker.
As I said many times before, TensorFold is a real turning point for local AI. Just try it for yourself, we have a recipe for everyone!
All of my TensorFold recipes can be found here:
Show more
@Beamsters1 It will go good with this commit right here ๐
i want my mac pro 2019 back :D
What a bonkers seven weeks it's been for Omarchy!
I use minimax h3 a lot itโs very good I set to build the fastest and best quality h3 implementation on mac with tensorfold . Coming to dgx sparks soon. Thank you @MiniMax_AI
Show more
Who's got the best open-source video world model when it comes to physics?๐ซ
it's me again, MiniMax H3. Almost on par with closed-source sota.๐พ
and..... Hermes install on Omarchy!
