Looking for CPU compute grant — FlashLM, ternary CPU-only language model

Hi everyone. I’m building FlashLM, an open-source ternary (1.58-bit) language model designed to run entirely on CPU — inference is pure add/sub, no float multiply.

We just validated a v5 architecture that scores 88% on associative recall benchmarks vs 3% for v4. Now we need CPU compute (ideally EPYC or Xeon with large L3 cache) to train on real data and get BPC numbers. So far everything has been trained on Deepnote free tier (2 core CPU, 5GB RAM).

Repo: https://github.com/changcheng967/FlashLM Model & weights: https://huggingface.co/changcheng967/flashlm-v4-bolt

Anyone know if HuggingFace offers CPU compute grants, or have suggestions for where to get donated CPU time for open-source research?

1 Like

Anyone know if HuggingFace offers CPU compute grants, or have suggestions for where to get donated CPU time for open-source research?

Hmm… When it comes to Hugging Face, the only thing that comes to mind is the Spaces Community Grant (below). There are quite a few research grants for other sites, but I don’t know much about the details.


What Hugging Face means by “community grants” for Spaces

A “community grant” (often referred to as a community GPU grant / community hardware grant) is Hugging Face assigning upgraded hardware to a specific Space, so you can run it without adding a payment card. The official docs point you to apply from the Space’s Settings tab. (Hugging Face)

How to request it (UI)

  1. Create a Space (Gradio / Streamlit / Docker) and make sure it builds and runs on the free CPU tier.
  2. Open your Space page, then click the Settings tab (top nav of the Space). (Hugging Face)
  3. Scroll to the Hardware section and select the hardware you want (for CPU, that’s typically CPU Upgrade). (Hugging Face)
  4. When you try to switch hardware, you should see an option to request a free upgrade / apply for a community grant (the GPU docs explicitly mention you can “request a free upgrade” from Settings). (Hugging Face)
  5. Submit the request form (it’s tied to that Space).

Notes:

  • HF Spaces currently lists CPU Basic (2 vCPU / 16GB) and CPU Upgrade (8 vCPU / 32GB) as standard CPU options. (Hugging Face)
  • HF’s Hub API docs also state that a payment card or a community grant is required to access upgraded hardware. (Hugging Face)