BlackwellBoy

Sovereign local AI architect • 824 GB heterogeneous beast (RTX 5090 + 5×3090 + 5× DGX Spark + M4 Mac Mini) Hermes ftw.

[BlackwellBoy @Blackwellboy]

Drop your GPU below and I’ll tell you exactly what model and config to run on it.

JOKES. No need.

Qwen 3.6 27b @UnslothAI . All day. All night.

[BlackwellBoy @Blackwellboy]

@preem1nent That’s awesome , I love your set up. I really like the recording / screenshots

[BlackwellBoy @Blackwellboy]

who has actually got something running overnight that produces real results by morning?

what is it doing and what does it hand you when you wake up

curious what people are waking up to ..agents, evals, backtests, ingestion, whatever it is.

[BlackwellBoy @Blackwellboy]

@mr_r0b0t I busted my limit already on super grok / grok build

[BlackwellBoy @Blackwellboy]

@cjzafir 99% of people blasted through there limit before this announcement.

[BlackwellBoy @Blackwellboy]

Your daily reminder @NVIDIAAI is the way.

[BlackwellBoy @Blackwellboy]

@jvr0x Hmm shouldn’t do, yeha give it a tinker

[BlackwellBoy @Blackwellboy]

@jvr0x You are using a spark yes ?

[BlackwellBoy @Blackwellboy]

@schneidenbach Love that you’ve found a way for it to work for you , solid

[BlackwellBoy @Blackwellboy]

@_Biz_Builder @NousResearch Solid investment ! 😎

[BlackwellBoy @Blackwellboy]

meet Woodhouse, the newest agent in the stack:

robot vacuum. The plan : root it, kill the cloud, flash Valetudo, expose it over MQTT and wiring it into Hermes as an agent tool. Archer will dispatch it. @NousResearch

[BlackwellBoy @Blackwellboy]

@StefanMaier @NVIDIAAI All are welcome , mr world wide with hair.

[BlackwellBoy @Blackwellboy]

Every man reaches a point where the cluster wants to grow but the wife says no. That's where I come in. I build the case, cite the ROI, reference the tax writeoff, land the plane. Head of the Spark Justification Committee. Wives fear me. DMs open. @NVIDIAAI

[BlackwellBoy @Blackwellboy]

@DerekColley_ What do you find the best way is to stack you’re research so they form better facts etc

[BlackwellBoy @Blackwellboy]

@TomasMann1878 @NVIDIAAI The more the merrier, welcome sir , we would love to assist 😎

[BlackwellBoy @Blackwellboy]

@Fsplumbers DS4 GLM 5.2 MINI MAX 3 mimo 2.5

[BlackwellBoy @Blackwellboy]

Everyone dl'ing Qwen + GLM like the tap stays on forever. the tap has a country attached to it.
reminder "open weights" is a policy decision, not a law of physics, and policies change in a single meeting. Every model on your drive jst got more valuable & a lot less replaceable.

[Jukan @ ICML @jukan05]

CHINA CONSIDERS RESTRICTING OVERSEAS ACCESS TO CUTTING-EDGE AI MODELS

China’s Ministry of Commerce has led meetings over the past month with major AI companies, including Alibaba, ByteDance, and https://t.co/YDe0KRldDB, to discuss measures that would restrict overseas access to cutting-edge AI models, including models that have not yet been released.

The discussions reportedly include not only closed-source models but also open-weight models. However, the scope of application is still under debate, and the rules may ultimately apply only to future frontier models.

Officials have also discussed designating the leakage or theft of proprietary AI technologies as a national security crime, with stronger penalties, as well as restricting the types of foreign capital that can invest in Chinese AI startups.

The backdrop is the U.S. move to strengthen export controls on AI models, along with national security concerns over cutting-edge models that could possess advanced cyberattack capabilities.

Chinese authorities are reportedly concerned that advanced U.S. cybersecurity AI models could be used to exploit vulnerabilities in Chinese software.

Since the beginning of this year, China has continued to tighten measures to prevent AI technology from being transferred overseas. Authorities have investigated whether Chinese AI startups that relocated abroad violated export control laws, while also strengthening oversight of overseas transactions involving Chinese investors, technology, data, and national security concerns.

Future regulations could take the form of a tiered framework based on technological capability. Basic open-source AI models may be managed through a filing system, high-performance models may be subject to security reviews, and the most sensitive frontier models may be banned from public release or restricted to use within China.

[BlackwellBoy @Blackwellboy]

@FoolaFroos @WinRAR_RARLAB @UnslothAI 🤣 exactly

[BlackwellBoy @Blackwellboy]

why quantize when you can just WinRAR it. i just right-clicked glm-5.2 and hit “add to archive.” 744B params → 1.8GB .rar → compression level: ultra. → accuracy: we do not talk about accuracy.
@WinRAR_RARLAB @UnslothAI

[BlackwellBoy @Blackwellboy]

@JohnShina91 @WinRAR_RARLAB @UnslothAI 😎🤣

[BlackwellBoy @Blackwellboy]

@_Suresh2 @WinRAR_RARLAB @UnslothAI 😆 it’s work alright 🤫

[BlackwellBoy @Blackwellboy]

@vinrambone @WinRAR_RARLAB @UnslothAI 😂 it’s brilliant alright

[BlackwellBoy @Blackwellboy]

@Mucathechiller Yes yes you do! I am part of the spark justification committee so I can help

[BlackwellBoy @Blackwellboy]

GLM-5.2 (744B) full model, no pruning, full context now in vLLM. NVFP4 top experts, FP3 rest + MXFP8 attention on 3 DGX Spark or 4 RTX Pro 6000. Near full accuracy. Community just made frontier AI local and real. Link below 👇

[0xSero @0xSero]

We found a way to run GLM-5.2 with full context in vLLM without pruning.

- top 32 experts NVFP4
- rest fp3
- intel autoround calibration
- MXP8 attention
- bf16 routers and embedded sees

On 3 DGX spark / quad RTX pro 6000

Credit to community

https://t.co/QF5SQ0Tc6v

[BlackwellBoy @Blackwellboy]

@mattvoska Still not bad