MLSys 2024 Papers

Skip to yearly menu bar Skip to main content

Layout:

mini compact topic detail

by

SiDA: Sparsity-Inspired Data-Aware Serving for Efficient and Scalable Large Mixture-of-Experts Models

Does Compressing Activations Help Model Parallel Training?

FlashDecoding++: Faster Large Language Model Inference with Asynchronization, Flat GEMM Optimization, and Heuristics

VIDUR: A LARGE-SCALE SIMULATION FRAMEWORK FOR LLM INFERENCE

CloudEval-YAML: A Practical Benchmark for Cloud Configuration Generation

LIFL: A Lightweight, Event-driven Serverless Platform for Federated Learning

FLASH: Fast Model Adaptation in ML-Centric Cloud Platforms

UniDM: A Unified Framework for Data Manipulation with Large Language Models

ACROBAT: Optimizing Auto-batching of Dynamic Deep Learning at Compile Time

FedTrans: Efficient Federated Learning via Multi-Model Transformation

JIT-Q: Just-in-time Quantization with Processing-In-Memory for Efficient ML Training

SLoRA: Scalable Serving of Thousands of LoRA Adapters

HeteroSwitch: Characterizing and Taming System-Induced Data Heterogeneity in Federated Learning

Fine-Tuning Language Models Using Formal Methods Feedback: A Use Case in Autonomous Systems

Distributed Matrix-Based Sampling for Graph Neural Network Training

Atom: Low-Bit Quantization for Efficient and Accurate LLM Serving

vMCU: Coordinated Memory Management and Kernel Optimization for DNN Inference on MCUs

AWQ: Activation-aware Weight Quantization for On-Device LLM Compression and Acceleration

VQPy: An Object-Oriented Approach to Modern Video Analytics

COMET: Neural Cost Model Explanation Framework

Schrodinger's FP Training Neural Networks with Dynamic Floating-Point Containers

Lancet: Accelerating Mixture-of-Experts Training by Overlapping Weight Gradient Computation and All-to-All Communication

Accelerating ReLU for MPC-Based Private Inference with a Communication-Efficient Sign Estimation

L-GreCo: Layerwise-adaptive Gradient Compression For Efficient Data-parallel Deep Learning

Keyformer: KV Cache reduction through key tokens selection for Efficient Generative Inference

Torch2Chip: An End-to-end Customizable Deep Neural Network Compression and Deployment Toolkit for Prototype Hardware Accelerator Design

ACCURATE LOW-DEGREE POLYNOMIAL APPROXIMATION OF NON-POLYNOMIAL OPERATORS FOR FAST PRIVATE INFERENCE IN HOMOMORPHIC ENCRYPTION

Prompt Cache: Modular Attention Reuse for Low-Latency Inference

Q-Hitter: A Better Token Oracle for Efficient LLM Inference via Sparse-Quantized KV Cache

Proteus: Preserving Model Confidentiality during Graph Optimizations

Efficient Post-training Quantization with FP8 Formats

On Latency Predictors for Neural Architecture Search

QMoE: Sub-1-Bit Compression of Trillion Parameter Models

DiffusionPipe: Training Large Diffusion Models with Efficient Pipelines

Punica: Multi-Tenant LoRA Serving

Disaggregated Multi-Tower: Topology-aware Modeling Technique for Efficient Large Scale Recommendation

HeteGen: Efficient Heterogeneous Parallel Inference for Large Language Models on Resource-Constrained Devices