SharQ: Bridging Activation Sparsity and FP4 Quantization for LLM Inference
cs.LG, cs.AI
Submitted: 2026-06-25
Updated: 2026-11-18
Code: https://github.com/actypedef/SharQ
Terminology
Sources
- QUIK: Towards End-to-End 4-Bit Inference on Generative Large Language Models
- PIQA: Reasoning about Physical Commonsense in Natural Language
- Oscillation-Reduced MXFP4 Training for Vision Transformers
- Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
- Four Over Six: More Accurate NVFP4 Quantization with Adaptive Block Scaling
- LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale
- Bridging the Gap Between Promise and Performance for Microscaling FP4 Quantization
- GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers
- FGMP: Fine-Grained Mixed-Precision Weight and Activation Quantization for Hardware-Accelerated LLM Inference
- GQA: A New Dataset for Real-World Visual Reasoning and Compositional Question Answering
- AMXFP4: Taming Activation Outliers with Asymmetric Microscaling Floating-Point for 4-bit LLM Inference
- SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models
- Evaluating Object Hallucination in Large Vision-Language Models
- MicroMix: Efficient Mixed-Precision Quantization with Microscaling Formats for Large Language Models
- SpinQuant: LLM quantization with learned rotations
- HiFloat4 Format for Language Model Inference
- ARCQuant: Boosting NVFP4 Quantization with Augmented Residual Channels for LLMs
- Pointer Sentinel Mixture Models
- Qwen2.5 Technical Report
- WinoGrande: An Adversarial Winograd Schema Challenge at Scale
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis