arXiv subjects

Ramasubramanian Balasubramanian

Publications and source records attributed to Ramasubramanian Balasubramanian.

4 recordsLinked to original sources

arXiv2026

ChunkWise LoRA: Adaptive Sequence Partitioning for Memory-Efficient Low-Rank Adaptation and Accelerated LLM Inference

Recent advances in low-rank adaptation (LoRA) have enabled efficient fine-tuning of large language models (LLMs) with minimal additional parameters. However, existing LoRA methods apply static rank configurations uniformly across all input tokens, ignoring variation in token complexity and computational requirements. In this work, we propose ChunkWise LoRA, a dynamic and adaptive approach that partitions sequences into variable-length chunks based on token complexity and assigns each chunk a tailored low-rank configuration. Our system introduces a runtime scheduler that estimates token difficulty, performs adaptive chunking, and selects per-chunk LoRA rank and scaling using a rank-ladder mechanism. To preserve output consistency, we further introduce a boundary-safe composition module and integrate policy-driven KV-cache strategies. Experiments on benchmark datasets such as Wikitext-103 and SQuAD demonstrate that ChunkWise LoRA achieves up to 34\% lower latency and 38% memory reduction compared to baseline LoRA, while maintaining or improving task performance metrics like BLEU, EM, and perplexity. The proposed framework remains fully compatible with existing transformer architectures and inference frameworks, providing a practical solution for real-world deployment of parameter-efficient LLMs.

cs.CL↗

arXiv2024

Effects of dislocation filtering layers on optical properties of third telecom window emitting InAs/InGaAlAs quantum dots grown on silicon substrates

Integrating light emitters based on III-V materials with silicon-based electronics is crucial for further increase in data transfer rates in communication systems since the indirect bandgap of silicon prevents its direct use as a light source. We investigate here InAs/InGaAlAs quantum dot (QD) structures grown directly on 5° off-cut Si substrate and emitting light at 1.5 micrometers, compatible with established telecom platform. Using different dislocation defects filtering layers, exploiting strained superlattices and supplementary QD layers, we mitigate the effects of lattice constant and thermal expansion mismatches between III-V materials and Si during growth. Complementary optical spectroscopy techniques, i.e. photoreflectance and temperature-, time- and polarization- resolved photoluminescence, allow us to determine the optical quality and application potential of the obtained structures by comparing them to a reference sample -- state-of-the-art QDs grown on InP. Experimental findings are supported by calculations of excitonic states and optical transitions by combining multiband k.p and configuration-interaction methods. We show that our design of structures prevents the generation of a considerable density of defects, as intended. The emission of Si-based structures appears to be much broader than for the reference dots, due to the creation of different QD populations which might be a disadvantage in particular laser applications, however, could be favourable for others, e.g. in broadly tunable devices, sensors, or optical amplifiers. Eventually, we identify the overall most promising combination of defect filtering layers, discuss its advantages and limitations, and prospects for further improvements.

cond-mat.mes-hall↗

arXiv2021

Variational Inference for Category Recommendation in E-Commerce platforms

Category recommendation for users on an e-Commerce platform is an important task as it dictates the flow of traffic through the website. It is therefore important to surface precise and diverse category recommendations to aid the users' journey through the platform and to help them discover new groups of items. An often understated part in category recommendation is users' proclivity to repeat purchases. The structure of this temporal behavior can be harvested for better category recommendations and in this work, we attempt to harness this through variational inference. Further, to enhance the variational inference based optimization, we initialize the optimizer at better starting points through the well known Metapath2Vec algorithm. We demonstrate our results on two real-world datasets and show that our model outperforms standard baseline methods.

cs.IR↗

arXiv2020

On Variational Inference for User Modeling in Attribute-Driven Collaborative Filtering

Recommender Systems have become an integral part of online e-Commerce platforms, driving customer engagement and revenue. Most popular recommender systems attempt to learn from users' past engagement data to understand behavioral traits of users and use that to predict future behavior. In this work, we present an approach to use causal inference to learn user-attribute affinities through temporal contexts. We formulate this objective as a Probabilistic Machine Learning problem and apply a variational inference based method to estimate the model parameters. We demonstrate the performance of the proposed method on the next attribute prediction task on two real world datasets and show that it outperforms standard baseline methods.

cs.LG↗