TokenRing: An Efficient Parallelism Framework for Infinite-Context LLMs via Bidirectional Communication
arxiv.org · 5,809 words · saved by 1 readers
N/A
TokenRing: An Efficient Parallelism Framework for Infinite-Context LLMs via Bidirectional Communication Zongwu Wang 1 2 Fangxin Liu* 1 2 Mingshuai Li 2 Li Jiang* 1 2 1. Department of Computer Science and Engineering, Shanghai Jiao Tong University, 2. Shanghai Qi Zhi Institute *Corresponding Author {wangzongwu, liufangxin, ljiang cs}@sjtu.edu.cn arXiv:2412.20501v1…
related reading
- Context Parallelism for Scalable Million-Token Inferencearxiv.org
- [2512.23675] End-to-End Test-Time Training for Long Contextarxiv.org
- LoongTrain: Efficient Training of Long-Sequence LLMs with Head-Context Parallelismarxiv.org
- Large Language Diffusion Modelsarxiv.org
- 2502.11089arxiv.org
- Overleaf Examplearxiv.org
- Transformers Inference Optimization Toolset | AstraBlogastralord.github.io
- Parallel CPU-GPU Execution for LLM Inference on Constrained GPUsarxiv.org
- 2310.01889arxiv.org
- Inference economics of language modelsarxiv.org
- [2605.06546] Efficient Pre-Training with Token Superpositionarxiv.org
- LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Modelsarxiv.org