PMRF Reviews

2024

PMRF Reviews

Recommended with commendation

Prateek presented his work “ Bayesian Coresets for Personalized Federated Learning” as part of his PhD topic “Towards building Efficient and Robust Methods in LLM Pretraining and Finetuning.” The presented work has been accepted in ICLR 2024 (A* conference). The extension of the ICLR work is submitted to AAAI. Also, another work “Few shot chain-of-thought driven reasoning to prompt LLMs for open-ended medical question answering,” accepted in EMNLP. In summary, the candidate has strong acceptance and submissions at A* conferences and contributes to the BharthGen ecosystem. Furthermore, he is clear about his future work and goals. Further, completed the required TA duties. However, the student must focus on the following: 1) Improving the bounds of the model with respect to the generic distance metric. 2) Motivation of the Hellinger distance is missing. 3) Presentation skill needs to be improved as the slides were full of text but lacked clarity.

2023

PMRF Reviews

Recommended with commendation: Prateek has done a thorough literature survey and identified interesting research directions. He has already submitted a paper in ICLR 2024 and received a positive feedback from the conference. The progress is commendable especially because Prateek is a direct PhD candidate. Expect Prateek to continue with the research plan and convert the starts into research publication outcome in the coming one year. The progress seems above satisfactory in terms of efforts and progress. The only reason committee does not recommend commendation (only given selectively) is lack of published articles during this period. The candidate should not be discouraged by this as he has made a significant progress.