Notes & comparisons on where to store large open-weight LLMs.
Ghi chú & so sánh các phương án lưu trữ mô hình open-weight dung lượng lớn.
⚠️ Work in progress / Đang trong quá trình nghiên cứu
- Local NVMe: fastest but limited capacity, no scalability.
- Google Cloud Storage (GCS): good for Vertex AI integration, per-GB cost.
- Amazon S3: widely supported, ecosystem mature.
- Hugging Face Hub: free hosting for smaller repos, community-friendly.
- [YYYY-MM-DD] Init repo.
- [YYYY-MM-DD] Added cost comparison notes.