Fangcheng Fu

Cascadia: An Efficient Cascade Serving System for Large Language Models

Quantization Meets Sparsification for Faster Image Generation

SALE : Low-bit Estimation for Efficient Sparse Attention in Long-context LLM Prefilling

NetMoE: Accelerating MoE Training through Dynamic Sample Placement

Demystifying Cost-Efficiency in LLM Serving over Heterogeneous GPUs

Enhancing Unsupervised Sentence Embeddings via Knowledge-Driven Data Augmentation and Gaussian-Decayed Contrastive Learning

Gradual Learning: Optimizing Fine-Tuning with Partially Mastered Knowledge in Large Language Models