San Francisco Tensor Company @SfTensor

turns out when you have too many GPUs for your own good you can do stupid stuff such as pre-training a foundation model with constrained decoding on PTX -- that is, it does not speak natural language, it's tokenizer is PTX, it's thoughts are PTX, it's all PTX 🧵

We're #1 on NVIDIA's SOL-ExecBench across all four tracks. Across 235 real production kernels run on B200s we beat out 583 teams, including teams that have raised hundreds of millions of dollars to optimize kernels. Along the way we learned a lot about agentic optimization 🧵

We've been working on training our own models for large-scale, long-running optimization tasks. The model is still training, but we asked an early checkpoint to optimize pre-training on 32 MI355x, surprisingly it manages to train faster than NVIDIA NeMo running on 32 GB200s 🧵

for RL and agents you need fast, isolated sandboxes which is why we've used gVisor for CPU and NVIDIA GPU workloads. as we started post-training models for AMD GPU kernels though, we hit a problem: gVisor has no AMD GPU support. or rather, it didn't, until we built it :)

We have Day‑0 support for Xiaomi-Robotics-1 at @SFTensor Let us help you specialize it for your use case

Xiaomi Tech @XiaomiTech_

3 months ago

Xiaomi-Robotics-1 is now open source!🎉 Xiaomi-Robotics-1 was pre-trained on more than 100,000 hours of UMI data and post-trained with over 10,000 hours of cross-embodiment data. This open-source release covers the complete pipeline from real-robot post-training to model

53 204 2K 133K 656

The biggest advantage of owning the backend end-to-end is that instead of relying on ptxas, you can try different ways of scheduling, such as search & RL. While ptxas beats our beta backend without search, once we turn on search we beat it by ~8 points 🧵

Ben Koska @BenKoska

3 months ago

we've long been dissatisfied with the fact that the closest we can get to the hardware on NVIDIA is PTX, with ptxas then making all sorts of decisions (instruction selection, barriers, stall counts, etc.) for us without any way to influence it, so we've been building a custom

15 7 138 12K 62

we've long been dissatisfied with the fact that the closest we can get to the hardware on NVIDIA is PTX, with ptxas then making all sorts of decisions (instruction selection, barriers, stall counts, etc.) for us without any way to influence it, so we've been building a custom LLVM backend that let's us create cubins directly from LLVM, which let's us do things such as write inline SASS and define custom scheduling. reply or reach out if you wanna give it a twirl

The benefits of AI are dangerously concentrated. Today they accrue to a handful of labs and to one industry, software, whose work happens to look like the internet the models were trained on. Changing that is the single biggest untapped lever for economic growth this decade.🧵

And because we automatically target whichever hardware is cheapest or best for a given workload, vendors have to compete on performance and price instead of coasting on lock-in. A chip that becomes uncompetitive can no longer hide behind switching costs. It has to win on the merits.

We built a software model of NVIDIA Blackwell's Tensor Core MMA instruction (tcgen05.mma) that matches real B200 hardware bit-for-bit. Not "close enough for numerics." Every output word, identical. Here's how we reverse-engineered it. 🧵

We've stopped testing our GPU kernels and started proving them correct instead. A test suite checks the inputs you thought of. A proof covers every input that exists and hands you the exact one that breaks the kernel if it's wrong. Here's how. 🧵

The moment we saw this letter, we signed. An open ecosystem beats a closed moat every time and is how AI actually reaches every sector. Proud to stand with everyone helping build and accelerate open, American AI

Brad Smith @BradSmi

3 months ago

Since launching last week, more than 230 companies and organizations from across the tech sector have signed the "Open Weights and American AI Leadership" open letter. We want to thank these partners for standing up and publicly supporting broader access to AI innovation. A

58 178 1K 359K 235

We built a benchmark where the AI agent has to build an inference chip. GPT-5.6 took 3.4 days and 1.7B tokens. The agent has to design, verify and physically place-and-route a working inference accelerator from a single MAC unit to a routed die that decodes tokens bit-exactly.

made it to Times Square! now back to building thanks @brexHQ for the support getting here

We're hiring a Founding Product Engineer. If we hire someone you refer, I'll wire you $15k. - Strong design taste - Experience with React/NextJS - Care deeply about user experience - Deep curiosity and drive to learn through experimenting 💵 $225k-275k base salary + equity

Excited to have powered the training infrastructure for @KalpaLabsAI's latest model. Congrats to @JhaGauti and @pshishodiaa on the launch and excited to see where you take it!

Y Combinator @ycombinator

11 months ago

Kalpa Labs (@KalpaLabsAI) is training one generalist model for all speech tasks. Steer it like an LLM, give long system prompts & make use of native contextual awareness. Congrats on the launch, @pshishodiaa and @jhagauti! ycombinator.com/launches/Op4-k…

22 36 269 97K 110

SF Tensor (@sftensor) is building infrastructure for training foundation models, letting researchers focus on research and forget that the infrastructure layer even exists while saving 70% in training costs. ycombinator.com/launches/Ojq-s… Congrats on the launch, @benkoska, @tom_koska, and @lukkoska!

One interface for everything. Automatically finds cheap GPUs, handles spot instances (saves up to 80%), moves jobs around when needed. Tell us what you need. Your training runs. That's it. cloud.sf-tensor.com

Running AI workloads across clouds is honestly a nightmare. GPU availability is random, prices swing wildly, and managing AWS + GCP + Azure + niche providers is basically a full-time job. We got sick of it, so we built Elastic Cloud. (1/2)