PhD Position F/M Reliability of Large Foundation Models

The Inria Rennes - Bretagne Atlantique Centre is one of Inria's eight centres and has more than thirty research teams. The Inria Center is a major and recognized player in the field of digital sciences. It is at the heart of a rich R&D and innovation ecosystem: highly innovative PMEs, large industrial groups, competitiveness clusters, research and higher education players, laboratories of excellence, technological research institute, etc.

Large Foundation Models (LFMs) are cutting-edge technology for natural language processing, object detection [1] and segmentation [1], and audio and multimodal processing, outperforming any available machine learning technique. LFMs, such as OpenAI GPT-4, Google ViT, and Meta LLaMA, have gained public attention with unprecedented accuracy. Given the superior performance of LFMs, they are being deployed in safety-critical and mission-critical applications, including space exploration [2] and self-driving cars. Improving LFMs’ reliability is crucial to enable dependable real-time safety-critical systems.

Large and complex accelerators like field-programmable gate arrays (FPGAs) are ideal for deploying LFMs in safety-critical applications [3]. However, LFM accelerators integrated into safety-critical systems must meet specific constraints,  including real-time execution and high classification/detection accuracy, even in harsh environments like space exploration [4]. It is imperative to evaluate whether these critical requirements are met when undesirable events, such as radiation-induced faults [4],  disrupt correct hardware execution and modify the expected results of the LFMs.

This Ph.D. aims to identify hardware and software vulnerabilities in LFM-based systems and propose error mitigation techniques.  

References

[1] Jakubik, Johannes et al. “Foundation Models for Generalist Geospatial Artificial Intelligence.” preprint, 2023

[2] Fang, Yuxin et al. “EVA-02: A Visual Representation for Neon Genesis.” preprint, 2023

[3] Knopp, T. et al. "AMD Versal™ AI Edge Series Gen 2 for Vision and Automotive." 2024 IEEE Hot Chips 36 Symposium, 2024.

[4] Maillard, Pierre, et al. "Heavy-ion and proton characterization of AMD 7nm Versal™ multicore scalar processing system (PS)." 22nd European Conference on Radiation and Its Effects on Components and Systems, 2022.

The Ph.D. student will characterize the impact of radiation-induced faults on system reliability on large accelerators like AMD Versal SoC for vision, language processing, and multimodal LFMs [3]. The results will be combined with software simulation data to identify effective hardening solutions. The Ph.D. student will work on developing new fault tolerance approaches tailored for LFMs. Standard fault tolerance techniques may introduce unacceptable overhead. We will conduct a comprehensive fault propagation analysis to propose efficient and effective hardening methods.
The Ph.D. student will participate in international experiments and internships at laboratories like Rutherford Appleton Laboratory in the  UK and Los Alamos National Laboratory in the USA. In addition, the student will participate in conferences and international projects. This can help them to develop their research skills and network with other professionals in their field.

Languages: proficiency in written English and fluency in spoken English are required.

Relational skills: the candidate will work in a research team, where regular meetings will be set up. The candidate has to be able to present the progress of their work in a clear and detailed manner.

Other values appreciated: Open-mindedness, strong integration skills, and team spirit.

Most importantly, we seek highly motivated candidates.

• Subsidized meals
• Partial reimbursement of public transport costs
• Possibility of teleworking (90 days per year) and flexible organization of working hours
• Partial payment of insurance costs

monthly gross salary amounting to 2100 euros for the first and second years and 2190 euros for the third year