Joey
@aijoey • 3,712 subscribers
Home AI lab: 2× DGX Spark · Mac Mini · RTX 4080 Hermes agents · Tools I ship · All things Open Source Dev Ambassador @Alibaba_Qwen
Shorts
Videos
Tired of seeing the same benchmarks around local models? I ran Poolside’s Laguna S 2.1 locally on a single NVIDIA DGX Spark. I gave it a simulated problem inside an electric vehicle battery factory. One of the machines kept producing parts with bad torque readings. The robot’s power usage and vibration were slowly increasing, but not enough to trigger any alarms. Maintenance thought the robot was failing. Production thought the factory was receiving defective parts. Meanwhile, more products were being rejected and the production line was starting to back up. Laguna reviewed the factory sensor data, worker notes, maintenance manuals, machine limits, quality records and the production schedule. It ruled out the incoming parts and found that several small warning signs pointed to a worn spindle bearing. Before changing anything, it tested a solution inside a digital factory simulation. It rerouted production to a backup machine, checked the results and then stopped to ask for human approval before continuing. The results: OEE improved from 71.4% to 93.1% Torque rejects dropped from 8.4% to 1.1% The production queue dropped from 18 packs to 5 The factory itself was simulated. The model run, tool calls and investigation were real and happened locally on one DGX Spark. This is not about letting AI control a factory by itself. It is about giving people another tool to understand complicated problems faster, connect signals they might normally miss and test possible solutions before touching the real production line. Really interesting work from Poolside, Eiso Kant and Jason Warner.
Joey11,000 görüntüleme • 17 gün önce
WeiboAI dropped VibeThinker-3B, so I had to try it locally. this is a 3B model, not a giant frontier system. in the video I load it on my DGX Spark, give it 3 small algorithm problems, stream the actual model output live, then run the generated python through pytest. no benchmark screenshot no canned answer just a tiny local reasoner writing code and real tests deciding if it worked it went 3/3.
Joey16,764 görüntüleme • 1 ay önce
got gemma 4 31B with MTP running on my DGX Spark. Hermes Agent did most of the legwork. baseline vs MTP on GB10: • c=1: 3.65 → 6.37 tok/s (1.74x) • c=4: 14.34 → 23.59 tok/s (1.65x) • c=8: 14.37 → 24.18 tok/s (1.68x) google says "up to 2x" — we're not quite there but it's real, not vapor. stack: DGX Spark / GB10 + gemma-4-31b-it + gemma-4-31b-it-assistant (MTP drafter) + vLLM built from PR 41745 MTP is basically a lightweight draft model that predicts multiple tokens while the big model verifies them all at once. smaller model does the busywork, bigger model just says yes/no. simple idea, weird to implement. next: tune the draft block size and see if we can push past 2x. also want to try it with Hermes Agent feeding prompts end to end. p.s: this was all done from telegram. Google DeepMind NVIDIA AI Developer
Joey23,088 görüntüleme • 3 ay önce
15 concurrent terminal workloads on a local DGX Spark, all served by nvidia/nemotron-3-super 120B A12B NVFP4 through vLLM. 15/15 completed, 0 errors, 30.2s wall time: no fake dashboard, just local inference under concurrent load. fineprint: live local concurrency demo
Joey11,375 görüntüleme • 2 ay önce
Daha fazla içerik yok.