Apply for community grant: Academic project (gpu)
Dear Hugging Face team,
We are a student team building a Schedule Organizer AI for our capstone project at Zaka AI. The system coordinates an agentic workflow that retrieves cohort, subgroup, and program data and generates conflict‑free schedules by following instructor‑defined rules. We use a retrieval‑augmented generation (RAG) pipeline to ground the model in our data and instructions.
GPU access would enable us to: (1) evaluate and serve open models (e.g., Llama 3 8B, Mistral 7B) with low latency, (2) generate embeddings efficiently at scale, and (3) run lightweight adapter fine‑tuning (PEFT/QLoRA) for better domain performance.
We respectfully request a temporary GPU grant for a Hugging Face Space: 1× T4 16GB or A10G 24GB for 3–4 weeks until the end of this October. We will host everything on the Hub: a public repo, a live Space demo, and clear model/dataset cards.
Deliverables:
Public Space with README and usage guide
Reproducible training/inference scripts
Evaluation summary (constraint satisfaction, accuracy, latency)
Short write‑up on approach, results, and lessons learned
Responsible AI: We work only with non‑sensitive academic data (no PII), and will document limitations and risk mitigations.
Thank you for considering our request and for supporting open‑source education. We’re happy to adapt to available hardware and share progress updates.
Best regards,
Rudolf Chakhtoura