Taking Tokenization off the Host for Agentic LLM Serving

arXiv:2610.13325 · cs.DC, cs.CL, cs.LG, cs.PF · Submitted 2026-10-09 · Read on arXiv

cs.DC, cs.CL, cs.LG, cs.PF

Submitted: 2026-10-09

Updated: 2026-10-09

Code: https://github.com/Msiavashi/GraphTok

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Tokenization — 1× in this paper · explained in 2 episodes

Sources

• GPUTOK: GPU Accelerated Byte Level BPE Tokenization
• OpenAssistant Conversations -- Democratizing Large Language Model Alignment
• The Llama 3 Herd of Models
• Qwen2.5 Technical Report
• Dissecting GPU Utilization for LLM Inference on Nvidia Hopper
• Fast WordPiece Tokenization
• TokTier: Exact Stateful CPU+GPU Tokenization for Agentic LLM Serving
• TraceLab: Characterizing Coding Agent Workloads for LLM Serving

Related papers

• iScheduler: Reinforcement Learning-Driven Continual Optimization for Large-Scale Resource Investment Problems
• SAMM: Sharded Automated Market Maker
• InferScale: GPU-Native KV Injection for Personalized LLM Serving
• Vigil: Accountable Liveness against Selective Silence
• Steelhead: Interleaving Partially Synchronous and Asynchronous Commit Rules on a Shared DAG
• Democratizing MoE inference on commodity GPUs with CoMoE