FAQ
Q1. For applicants affiliated with an academic institution, is professorial status required? Can graduate or doctoral students submit applications independently?
Only professors from academic institutions may serve as the project’s Principal Investigator.
Q2. In addition to completing the online application form, is a project proposal also required?
Teams that have a complete project proposal (in any format) may upload it during the online submission. If no proposal is available, the online form by itself will be accepted.
Q3. Are standardized formats or electronic templates available for both the project proposal and the project final outcome report?
No specific formats or templates are required for the project proposal or final outcome report; teams may prepare their documents independently.
Q4. If applicants are simultaneously applying for other computing subsidy programs, are they still eligible to apply for this program?
Yes; however, applicants are advised to disclose this information truthfully on the application form.
Q5. What resources will be provided to the selected teams?
This program provides GPU computing hours, storage space, and an orientation on using the GMI platform. Applicants are responsible for preparing their own models and data and must possess the technical skills necessary for data analysis, model training, MLOps deployment, multi-GPU operations, cloud environment deployment, code debugging, and other related capabilities.
Q6. Will the program provide models, or are teams required to develop their own?
The program does not provide models; participants must upload their own models for training or inference.
Q7. If we apply for eight GPUs, can the computing service be provided in Bare Metal mode?
Due to current functional limitations of the GMI platform, this program provides computing services only in Container mode. If there is significant demand for Bare Metal mode, the organizers will explore feasible solutions in the future.
Q8. If I am only familiar with Colab or on-premises systems and not with Docker or SSH, will training be offered to run my code on the program’s computing platform?
Teams requiring technical guidance on cloud environment deployment and operations are encouraged to consult directly with IT service providers for project support.
Q9. If the allocated GPU hours are insufficient, can additional GPU hours be requested?
If allocated GPU hours run out before the end of the usage period, teams may request a one-time supplement by submitting an application at least six weeks before the period ends. Approved extra hours must be used within the original project period, and any unused hours will be forfeited. The usage period cannot be extended.
Q10. Can the usage period be extended if allocated GPU hours are not fully used?
If allocated GPU hours are not fully used within the usage period and an extension is needed, each project may apply for a one-time extension. Applications must be submitted at least six weeks before the usage period ends. Once approved, the usage period may be extended by up to three months. Any unused GPU hours after the extended deadline will be forfeited, and no further supplementation will be granted.
Q11. What differentiates this program from other GPU resource programs?
1. Target Audience: This program focuses on medium-scale computing needs (1–8 H100 GPUs).
2.Duration of Computing Resources: This program provides computing resources for six months.
3.Project Outcome: Upon completion of the computing period, this program will host an annual results presentation to showcase project achievements.
4.Rooted in Taiwan: This program primarily targets startups with local connections in Taiwan and Taiwanese universities, with the goal of attracting foreign startup teams to establish and grow in Taiwan.
