影响力指数
论文质量、代表作、近期表现、广度与样本量置信度综合计算
71.68/100
前 2.3%
全站排名 #1,506
发表论文19 篇
平均评分
年均产出6.3 篇/年
Souvik Kundu
研究方向
Mixture of Agents · Reasoning Models · Inference Efficiency and Optimizations for LLMs · Long context Understanding and Reasoning of LLMs · Mixture of Experts · Pruning of DNN Model for Computer vision and NLP · Federated Learning for Resource-constrained Edge · Model and Data Privacy in Deep Learning · Brain-Inspired Spiking Neural Network Training and Compression Algorithms · Efficient Training of Robust DNNs against Adversarial Attacks · Machine Learning Architectures
21
PRISM: Enhancing PRotein Inverse Folding through Fine- Grained Retrieval on Structure-Sequence Multimodal Representations
ICLR 2026Poster
二作4
CLEAR: A Cost-Aware Routing System for Edge-Cloud Language Model Collaborative Inference
ICLR 2026Desk Rejected
17
MAPLE: Masked Adapter Prototype Learning for OOD generalization
ICLR 2026Rejected
25
CITER: Collaborative Inference for Efficient Large Language Model Decoding with Token-Level Routing
COLM 2025Poster
11
On-the-Fly Adaptive Distillation of Transformer to Dual-State Linear Attention for Long-Context LLM Serving
ICML 2025Poster
三作17
SEAL: Steerable Reasoning Calibration of Large Language Models for Free
COLM 2025Poster
24
LANTERN: Accelerating Visual Autoregressive Models with Relaxed Speculative Decoding
ICLR 2025Poster
24
Top-H Decoding: Adapting the Creativity and Coherence with Bounded Entropy in Text Generation
NeurIPS 2025Poster
三作6
MambaExtend: A Training-Free Approach to Improve Long Context Extension of Mamba
ICLR 2025Poster
二作22
Scaling Long Context Training Data by Long-Distance Referrals
ICLR 2025Poster
16
VERT: A SystemVerilog Assertion Dataset to Improve Hardware Verification with LLMs
ICLR 2025Rejected
6
CITER: Collaborative Inference for Efficient Large Language Model Decoding with Token-Level Routing
ICLR 2025Rejected
5
MeZO-A$^{3}$dam: Memory-efficient Zeroth-order Adam with Adaptivity Adjustments for Fine-tuning LLMs
ICLR 2025Withdrawn
7
ShareLoRA: Less Tuning, More Performance for LoRA Fine-tuning of LLMs
ICLR 2025Withdrawn