Posts by Guihong Li
Zebra-HyLo: Upcycling Transformers into Long-Context Hybrid LLMs on AMD Instinct™ GPUs
- 22 September 2026
Hybrid language models that interleave attention with linear sequence-modeling blocks have become the default answer to the cost of long context. Jamba, Samba, Qwen3-Next and Kimi-Linear all take this shape, and they all share one property that makes them expensive to adopt: they are pretrained from scratch. Every one of them pays the full cost of building a foundation model again, which means the enormous investment already sunk into existing Transformer checkpoints is thrown away.
Accelerating Mixture-of-Experts Execution with FarSkip-Collective Models
- 05 May 2026
Whether you are running training or inference, the largest Mixture-of-Experts (MoE) based LLMs cannot fit on a single GPU; instead you must run collective-communication operations to integrate the work of multiple GPUs to work together on a single model.
AMD-HybridLM: Towards Extremely Efficient Hybrid Language Models
- 17 September 2025
The rapid rise of deep learning applications has intensified the demand for language models that offer a balance between accuracy and efficiency—especially in settings constrained by memory, compute, or real-time requirements. While Transformer-based models have revolutionized natural language processing, their quadratic attention complexity and large key–value (KV) cache requirements pose serious challenges for deployment, particularly on edge devices or in latency-sensitive environments.