Skip to content
AI Atlas
Research

Papers

Publications linked to models, labs and benchmarks. Authors, venues and abstracts come from arXiv and publisher pages.

10 total

Reset
Papers
TitleAuthorsPublishedCategoriesOrganization / venueQuality
EMMI: Edge Multi-Modal Intelligence for Communication-Efficient MLLM Inference via Fused Representation CompressionarXiv:2609.11058Motahare Mounesan, Irfan Khan11 Sept 2026cs.LG—
Phase-Decoupled, Model-Calibrated Power Control for Disaggregated LLM ServingarXiv:2609.11133Jae Gon Kim, Donghoon Yoo, Hanyul Ryu +211 Sept 2026cs.LG—
MUC-FL: Block-Wise Marginal Utility Contribution for Communication-Efficient Federated LearningarXiv:2609.10545Akshay Mhatre, Vikram Karthick, Deepti Gupta +111 Sept 2026cs.DC—
Composable CXL Memory as a Kubernetes-Native Shared Memory for LLM ServingarXiv:2609.10790Hongjian Fan, Kevin Zhang, David Habinsky +111 Sept 2026cs.DC—
Building py-kvcache: A Performance Characterization of External KV Caching for vLLM with NVMe SSDsarXiv:2609.11744Joseph Kanichai, Tiziano De Matteis, Animesh Trivedi11 Sept 2026cs.DC—
SparseDitto: An Agentic Sparse Compilation Framework through Architecture-Aware Synthesis on GPUsarXiv:2608.05033Shiyang Li, Guangyan Sun, Jinwei Tang +211 Sept 2026cs.DC—
Smart Adaptive Computing Across the Continuum: LLMs in IoT-Edge-Cloud Resource ManagementarXiv:2609.09348Antonino Vaccarella, Lanpei Li, Vincenzo Lomonaco +111 Sept 2026cs.DC—
Influence-Oriented Personalized Federated LearningarXiv:2410.03315Yue Tan, Guodong Long, Jing Jiang +111 Sept 2026cs.LG—
FP8 is All You Need (Part 1): Debunking Hardware FP64 as the HPC Holy Grail (Sep 3rd version)arXiv:2606.06510Satoshi Matsuoka11 Sept 2026cs.AR—
FP8 is All You Need (Part 2): Full-FP64 3-D FFT on FP8-Generation Tensor CoresThe Integer-Epilogue Wall and the Minimal Hardware That Would Remove ItarXiv:2606.23698Satoshi Matsuoka11 Sept 2026cs.MS—

10 results

Author lists and categories are copied from the paper's own metadata (arXiv, publisher). Each paper page shows the abstract, related models and every source snapshot.