arXiv subjects

Sethuraman Sankaran

Publications and source records attributed to Sethuraman Sankaran.

3 recordsLinked to original sources

arXiv2026

Agentic Framework for Deep Learning workload migration via In-Context Learning

Translating deep learning models from PyTorch's flexible, object-oriented design to JAX's functional, stateless setup is usually a manual and error-prone task. Automated migration is challenging because Large Language Models (LLMs) struggle with strict and dynamic API alignment and are prone to mistakes for exacting operations. We propose a fully autonomous system that combines In-Context Learning (ICL) with oracle-driven self-debugging. First, we curated an ICL context that serves as a strict reference for idiomatic JAX styling and test case generation. Second, instead of depending on the LLM to deduce mathematical outputs, we run the source PyTorch modules to get their actual dynamic tensor states. This creates an unchangeable execution oracle. We then use an autonomous agentic loop to synthesize tests based on the oracle data. The test cases are executed repeatedly, and the traceback is sent back to the LLM for self-correction. Ablations show that combining ICL references with oracle grounding and self-debugging greatly outperforms pure instructional and basic agentic baselines. This improvement does not add an excessive computational overhead. Our lightweight pipeline achieves 91% numerical equivalence (compared to baseline: 9%, instruction + self-debugging: 27%) on neural modules, providing a highly reliable, scalable blueprint for cross-framework migration. This has been validated across several state-of-the-art models including SAM (segment anything), T5, Code Whisper amongst others showing high numerical equivalency. Code: https://github.com/AI-Hypercomputer/accelerator-agents/tree/main/MaxCode

cs.AI↗

arXiv2021

Multimodal Fusion Refiner Networks

Tasks that rely on multi-modal information typically include a fusion module that combines information from different modalities. In this work, we develop a Refiner Fusion Network (ReFNet) that enables fusion modules to combine strong unimodal representation with strong multimodal representations. ReFNet combines the fusion network with a decoding/defusing module, which imposes a modality-centric responsibility condition. This approach addresses a big gap in existing multimodal fusion frameworks by ensuring that both unimodal and fused representations are strongly encoded in the latent fusion space. We demonstrate that the Refiner Fusion Network can improve upon performance of powerful baseline fusion modules such as multimodal transformers. The refiner network enables inducing graphical representations of the fused embeddings in the latent space, which we prove under certain conditions and is supported by strong empirical results in the numerical experiments. These graph structures are further strengthened by combining the ReFNet with a Multi-Similarity contrastive loss function. The modular nature of Refiner Fusion Network lends itself to be combined with different fusion architectures easily, and in addition, the refiner step can be applied for pre-training on unlabeled datasets, thus leveraging unsupervised data towards improving performance. We demonstrate the power of Refiner Fusion Networks on three datasets, and further show that they can maintain performance with only a small fraction of labeled data.

cs.CV↗

arXiv2019

Physics driven reduced order model for real time blood flow simulations

Predictive modeling of blood flow and pressure have numerous applications ranging from non-invasive assessment of functional significance of disease to planning invasive procedures. While several such predictive modeling techniques have been proposed, their use in the clinic has been limited due in part to the significant time required to perform virtual interventions and compute the resultant changes in hemodynamic conditions. We propose a fast hemodynamic assessment method based on first constructing an exploration space of geometries, tailored to each patient, and subsequently building a physics driven reduced order model in this space. We demonstrate that this method can predict fractional flow reserve derived from coronary computed tomography angiography in response to changes to a patient-specific lumen geometry in real time while achieving high accuracy when compared to computational fluid dynamics simulations. We validated this method on over 1300 patients that received a coronary CT scan and demonstrated a correlation coefficient of 0.98 with an error of 0.005 +- 0.015 (95% confidence interval: (-0.020, 0.031)) as compared to three-dimensional blood flow calculations.

math.NA↗